Graduate Theses and Dissertations
Graduate College
2015
A manipulation of pyrolysis reaction pathways for
selective bio-oil composition
Yongsuck Choi
Iowa State University
Follow this and additional works at: http://lib.dr.iastate.edu/etd
Part of the Oil, Gas, and Energy Commons
Recommended Citation
Choi, Yongsuck, "A manipulation of pyrolysis reaction pathways for selective bio-oil composition" (2015). Graduate Theses and
Dissertations. Paper 14300.
This Dissertation is brought to you for free and open access by the Graduate College at Digital Repository @ Iowa State University. It has been accepted
for inclusion in Graduate Theses and Dissertations by an authorized administrator of Digital Repository @ Iowa State University. For more
information, please contact [email protected].
A manipulation of pyrolysis reaction pathways for selective bio-oil composition
by
Yongsuck Choi
A dissertation submitted to the graduate faculty
in partial fulfillment of the requirements for the degree of
DOCTOR OF PHILOSOPHY
Major: Chemical Engineering
Program of Study Committee:
Brent H. Shanks, Major Professor
Robert C. Brown
Jean-Philippe Tessonnier
Jacek A. Koziel
Terrence R. Meyer
Iowa State University
Ames, Iowa
2015
Copyright © Yongsuck Choi, 2015. All rights reserved.
ii
DEDICATION
I would like to dedicate my dissertation work to my beloved wife, Kyung Eun Lee,
who has encouraged me to do my best throughout the course of this research.
iii
TABLE OF CONTENTS
Page
DEDICATION ...........................................................................................................
ii
ACKNOWLEDGMENTS .........................................................................................
iv
ABSTRACT………………………………. ..............................................................
v
CHAPTER 1
GENERAL INTRODUCTION .......................................................
1
CHAPTER 2 PROPOSED PYROLYSIS REACTION NETWORKS FOR
LIGNIN MODEL COMPOUNDS WITH β-O-4 AND α-O-4 LINKAGE ...............
4
CHAPTER 3 DETAILED CHARACTERIZATION OF RED OAK-DERIVED
PYROLYSIS OIL; INTEGRATED USE OF GC, HPLC, IC, GPC AND KARL
FISCHER…….. .........................................................................................................
39
CHAPTER 4 MANIPULATION OF CHEMICAL SPECIES IN BIO-OIL
USING IN SITU CATALYTIC FAST PYROLYSIS IN BOTH BENCH-SCALE
FLUIDIZED BED PYROLYZER AND MICROPYROLYZER ..............................
70
CHAPTER 5 ACETIC ACID REMOVAL FROM PYROLYSIS VAPORS
USING CaCO3 …………………………………………………………………… 101
CHAPTER 6
GENERAL CONCLUSION……………………………………... 136
iv
ACKNOWLEDGMENTS
I would like to thank my major professor, Dr. Brent Shanks for his educational
guidance and support throughout the course of this research.
In addition, I would also like to thank my committee members, Dr. Robert Brown,
Dr. Jean-Philippe Tessonnier, Dr. Jacek Koziel, and Dr. Terrence Meyer
Finally, thanks to my family for their encouragement and to my wife for her hours of
patience, respect, and love.
v
ABSTRACT
Fast pyrolysis is a promising biomass conversion technology to convert solid biomass
to liquid bio-oil that has a potential to be upgraded into chemicals or fuels. However, high
acidity and reactivity of a raw bio-oil creates utilization problems such as instability and
corrosivity during storage, making the oil less attractive as renewable sources. Thus,
knowledge of how to manipulate bio-oil composition by reducing acidity or reactivity during
pyrolysis is important in commercializing pyrolysis process. Research in this dissertation
focuses on a manipulation of the bio-oil species with different approaches.
While significant effort has been spent to understand the pyrolysis chemistry of
cellulose with key success, a complex structure of lignin has been a bottleneck in
understanding lignin pyrolysis. Thus, pyrolysis reaction pathways of lignin model
compounds have been investigated in attempt to understand native lignin pyrolysis. This
study suggests that not only free radical reaction which has been predominantly proposed in
literatures but also pericyclic reaction is significantly involved in pyrolysis.
Secondly, catalytic manipulation of bio-oil composition was quantitatively studied by
performing a full mass balance on all of the products generated from in situ CFP. This study
demonstrates that the high consumptions of C and H of bio-oil during CFP led to lower
energy recovery with CFP than thermal pyrolysis.
Due to the low energy recovery from in situ CFP, a third study aimed to study a vapor
manipulation using ex situ adsorbent. CaCO3, adsorbent, showed a high activity for acetic
acid removal by forming calcium acetate. It was also found that selectivity toward acid
vi
removal was improved by a partial regeneration of the material to lead to a presence of the
relatively inert acetate group on surface.
1
CHAPTER 1
GENERAL INTRODUCTION
Introduction
The United States Energy Independence and Security Act (EISA) was made on 2007,
aiming to replace fossil fuels with clean, sustainable, renewable fuels. Thus, research in
renewable, sustainable energy has recently grown, such as solar, wind, geothermal, and biomass
energy. Among the renewable sources, biomass has gained a great attention due to its
inexpensiveness and abundance[1]. Biomass is composed of cellulose, hemicellulose, lignin,
mineral and extractives. Cellulose is a polysaccharide of glucose unit linked glycosidic bonds,
and hemicellulose is a hetero-polysaccharide of pentose and hexose. A third component, lignin,
consists of a phenylpropane monomer connected by C-O and C-C linkages. The main
components are known to be chemically bonded such as hydrogen bond and covalent bond[2].
There are two main routes for biomass conversion, biochemical and thermochemical conversion
technologies[3]. Biochemical conversion uses microorganisms or enzyme, while thermochemical
conversion uses heat and catalysts. Thermochemical conversion has an advantage over the other
route, in that a whole biomass was utilized in thermochemical conversion, while biochemical
conversion uses only carbohydrate portions leaving lignin as a process heat. Among
thermochemical conversions (combustion, gasification and pyrolysis), a fast pyrolysis is
considered as one of the most promising thermochemical technologies, due to converting solid
biomass to liquid bio-oil. Fast pyrolysis is defined as a thermal depolymerization in absence of
oxygen at 400-600 °C, producing bio-oil as the desired product, along with char and noncondensable gases as by-products. Since the bio-oil has over 300 organic compounds[4, 5], it has
a high potential to be chemically upgraded into chemicals or fuels. However, bio-oil has large
2
amount of acidic and unstable compounds, which makes the oil less attractive as renewable fuels.
Extensive studies on catalytic upgrading of the bio-oil have been performed to control the bio-oil
species, such as steam reforming and hydrocracking[6-8]. Recently, a manipulation in vapor
phase has been investigated using CFP, among which zeolite material has been commonly used
as catalyst to generate aromatics[9, 10]. A solid char has been used as a soil enhancer to achieve
carbon-negative biofuels, since its high surface area can efficiently keep water and nutrients in
soil[2].
Problem identification
Since each catalytic process prefers a different bio-oil composition, a raw bio-oil having complex
composition cannot be upgraded by a single catalytic process. In fact, bio-oil has over 300
organic compounds and typically classified into low-molecular weight compounds, organic
acids, furan/pyran derivative compounds, phenolic compounds and anhydrosugars. Thus, a
manipulation of bio-oil composition is necessary prior to a specific upgrading method. In situ
CFP with zeolites has been used to generate aromatic hydrocarbons; however, a full
understanding of the manipulation has not been made due to an absence of full product
speciations. In the case of lignin, reaction pathways that govern product speciations have not
been fully understood, due to its complex structure. This suggests that lignin pyrolysis pathways
need to be investigated for its efficient upgrading.
Dissertation organization
Work in this dissertation is summarized in four chapters as well as general introduction and
general conclusion. Chapter 2 focuses on the study of reaction pathways of lignin model
compounds. Product distribution as well as rate constant calculation was used to propose a
reaction network for the compounds. The pathways provide a mechanistic insight to native lignin
3
pyrolysis. Chapter 3 is dedicated to a full characterization of bio-oil with an integrated use of
GC, HPLC, GPC, and IC. Chapter 4 focuses on the investigation of catalytic manipulation of
bio-oil composition using in situ CFP. Full mass balance allowed to fully understanding the
effects of acidity or basicity to the final bio-oil composition. Chapter 5 is dedicated on the study
of selective acid removal from pyrolysis vapors using CaCO3. Insights to acid removal and
material regeneration were provided by the extensive characterization of postreaction material.
References
1.
Huber, G.W., S. Iborra, and A. Corma, Synthesis of transportation fuels from biomass:
chemistry, catalysts, and engineering. Chemical reviews, 2006. 106(9): p. 4044-4098.
2.
Mohan, D., C.U. Pittman, and P.H. Steele, Pyrolysis of Wood/Biomass for Bio-oil: A
Critical Review. Energy & Fuels, 2006. 20(3): p. 848-889.
3.
Stevens, C. and R.C. Brown, Thermochemical processing of biomass: conversion into
fuels, chemicals and power. Vol. 12. 2011: John Wiley & Sons.
4.
Choi, Y.S., et al., Detailed Characterization of Red Oak-Derived Pyrolysis Oil;
Integrated use of GC, HPLC, IC, GPC and Karl-Fischer. Journal of Analytical and
Applied Pyrolysis, 2014.
5.
Mullen, C.A. and A.A. Boateng, Chemical Composition of Bio-oils Produced by Fast
Pyrolysis of Two Energy Crops†. Energy & fuels, 2008. 22(3): p. 2104-2109.
6.
Baker, E. and D. Elliott, Catalytic upgrading of biomass pyrolysis oils, in Research in
thermochemical biomass conversion1988, Springer. p. 883-895.
7.
Huber, G.W., R.D. Cortright, and J.A. Dumesic, Renewable Alkanes by Aqueous‐Phase
Reforming of Biomass‐Derived Oxygenates. Angewandte Chemie International Edition,
2004. 43(12): p. 1549-1551.
8.
Huber, G.W., et al., Production of liquid alkanes by aqueous-phase processing of
biomass-derived carbohydrates. Science, 2005. 308: p. 1446-2079.
9.
Carlson, T.R., et al., Aromatic production from catalytic fast pyrolysis of biomass-derived
feedstocks. Topics in Catalysis, 2009. 52(3): p. 241-252.
10.
Carlson, T.R., T.P. Vispute, and G.W. Huber, Green gasoline by catalytic fast pyrolysis
of solid biomass derived compounds. ChemSusChem, 2008. 1(5): p. 397-400.
4
CHAPTER 2
PROPOSED PYROLYSIS REACTION NETWORKS FOR LIGNIN MODEL COMPOUNDS
WITH β-O-4 AND α-O-4 LINKAGE
Yong S. Choi, and Brent H. Shanks
Abstract
The pyrolysis of seven lignin model compounds (five β-O-4 and two α-O-4 linked
molecules) was performed in a micropyrolyzer (at 500 °C) directly connected to GC-MS/FID, in
an attempt to determine governing reaction pathways. According to the mole balance throughout
the reaction, concerted retro-ene fragmentation and homolytic dissociation are strongly
suggested as an initial reaction step for β-O-4 compounds and α-O-4 compounds, respectively.
The difference in reaction pathway between compounds with different linkages is believed to
come from thermodynamics of radical initiation (by homolysis). Ab-initio calculations using
density function theory (B3LYP) were conducted to estimate the rate constants for the different
reaction pathways. The experiment results are well supported by the calculated rate constants.
Surprisingly, the suggested reaction pathways for simpler model compounds were able to
describe a pyrolysis of a trimeric lignin model compound containing both β-O-4 and α-O-4
linkages. Due to its structural similarity to native lignin, the reaction pathways for trimer
pyrolysis can be served as a building block for understanding pyrolysis chemistry of native
lignin.
Introduction
Interests in liquid biofuel have recently grown due to the depletion of fossil resources,
increasing energy demand, and concerns over global warming. Although ethanol is produced by
a biochemical conversion of cellulosic biomass, the conversion technology makes use of
5
cellulose and hemicellulose, leaving lignin as a low-grade process heat or waste[1-4]. Lignin is
the second most abundant component in biomass, and paper and pulp industries generate a huge
amount of lignin as a by-product. Thus, for biorefinery to be cost-competitive over petroleum
refinery, lignin utilization is imperative[5]. In fact, lignin has a huge potential to produce valueadded chemicals such as benzene, toluene and xylene, due to its aromatic structure. Fast
pyrolysis is a promising thermochemical technology for converting lignin into the chemicals[6,
7]. Fast pyrolysis is a rapid thermal decomposition of biomass to generate bio-oil, char and
gases, carried out at approximately 500 °C, in the absence of oxygen, and at atmospheric
pressure[6, 8]. Thus, understanding of how lignin is thermally degraded will be important in
designing and optimizing pyrolysis processes toward commercial production of chemicals.
Unlike the relatively homogeneous structure of cellulose (a polysaccharide of glucose)
and hemicellulose (a polysaccharide of glucose, mannose, galactose, xylose and arabinose)[8],
lignin is an amorphous polymer consisting of phenylpropane monomer with hydroxy and
methoxy groups on a ring[9, 10]. Depending on the number of methoxy group attached, the
phenylpropane monomer is classified into three basic units: 4-hydroxyphenylpropane, 4hydroxy-3-methoxy phenylpropane (guaiacylpropane), and 4-hydroxy-3,5-dimethoxy
phenylpropane (syringylpropane). These units are generally denoted as H, G, and S, respectively.
Proportion of the units in lignin varies with biomass source. For example, spruce lignin
(softwood) predominantly contains G unit (around 90 mol %)[11, 12], beech lignin (hardwood)
contains S (63 mol %) and G (37 mol %) units[11], and corn stover lignin (herbaceous)[12]
contains roughly equal amounts of H (36 mol %), G (34 mol %), and S (30 mol %) units. As
shown in Figure 1, these units in a native lignin are linked by C-O and C-C bonds: mainly β-O-4,
α-O-4, 4-O-5, 5-5, β-5, and β-β.[13] Among the linkages, C-O aryl ether bonds including β-O-4,
6
α-O-4 and 4-O-5 predominantly account for 58 % (softwood) and 74.5 % (hardwood) of total
linkages in lignin[9, 10]. In addition to complex structure of lignin, lignin is intimately connected
to carbohydrates by hydrogen bond and covalent bond in biomass. As a result, lignin isolation
from biomass without its structure modification is difficult with currently available methods[1].
In fact, it was reported that cleavage of β-O-4 linkage occurred during a milling process
generating a milled wood lignin, which is considered as the most representative of native
lignin[14]. Structural modifications upon isolation were such large that the pyrolysis behavior of
lignin varies with lignin isolation methods[15-17]. While significant effort has been spent to
understand the pyrolysis chemistry of cellulose with key success, the complex structure of lignin
and structural modification of lignin upon isolation have been a bottleneck in understanding
lignin pyrolysis. To overcome the issues, fast pyrolysis of lignin model compounds has been
studied. The lignin model compounds have a simple structure, typically a phenolic dimer linked
by β-O-4 or α-O-4 ether bond that is the most abundant linkage in lignin. However, most of these
studies were performed in experimental setups, which cannot meet a requirement of a typical fast
pyrolysis: high temperature of 500 °C, high heating rate of over 1000 °C/s, short vapor residence
time of less than 20 ms, and rapid cooling of volatiles. For example, pyrolysis of 2-phenethyl
phenyl ether (PhCH2 CH2OPh: β-O-4 linkage) was carried out in a Pyrex tube placed onto
fluidized sand bath[18]. In the experiment, not only an insufficient heating rate but also a
preheating step of the vessel at 120 °C for 5 minutes can lead to reactions not typically occurring
during fast pyrolysis. An inability of the experimental setups in providing a fast pyrolysis
condition makes it problematic to fundamentally understand pyrolysis chemistry.
There have been two main proposed reaction mechanisms for pyrolysis of lignin model
compounds. These include a free-radical reaction and concerted pericyclic reaction. In
7
literatures, a free-radical reaction was dominantly proposed, in which radicals are generated by a
homolytic cleavage of a weak aryl C-O bond and the radicals are involved in hydrogen
abstraction, dimerization and oligomerization[18-22]. Recently, β-O-4 type oligomeric model
compound, which is structurally closer to native lignin than simple phenolic dimers, was
pyrolyzed in the temperature range of 250-550 °C in a Pyroprobe and they proposed radical
reactions to account for the experimental observations[22]. However, the proposed reaction
mechanism may not be correct since the mechanism was not based on product distributions with
their relative proportions, but largely based on product identification. In general, a relative
amount of each product as well as product identifications is important in determining a reaction
mechanism[23]. On the other hand, concerted pericyclic reaction is another reaction pathway to
describe pyrolysis of lignin model compounds. Klein et al. performed a pyrolysis of 2-phenethyl
phenyl ether (PE), β-O-4 lignin model compound, at temperatures from 300 to 500 °C in the
presence and absence of tetralin, a radical scavenger, and calculated kinetic parameters, preexponential factor and activation energy[24]. Based on no significant changes in product
distributions between neat pyrolysis and co-pyrolysis with tetralin, they concluded that
predominant reaction to occur during pyrolysis of the β-O-4 compound is not free-radical
reaction but concerted retro-ene fragmentation. Recently, more detailed study of the same
compound was conducted by Jarvis et al. based on experimental and computational works[25].
They pyrolyzed the model compound in a hyperthermal nozzle connected to photoionization
time-of-flight mass spectrometry (PIMS) to allow short residence time of 100 µs and direct
detection of initial reaction products including intermediates and radicals. Experimental
observation and rate constant calculations suggested the concerted retro-ene and Maccoll
reactions were a dominant reaction at typical pyrolysis temperature between 500 and 600 °C,
8
while C-O homolytic bond scission reaction becomes dominant at higher temperatures (over
1000 °C).
In this study, aryl ether model compounds linked by the most common linkages in lignin
such as β-O-4 and α-O-4 were selected and pyrolyzed in a micropyrolyzer directly connected to
GC-MS/FID. The simplest β-O-4 compound is 2-phenoxyphenylethanol (PPE:
PhCHOHCH2OPh). To characterize the impacts of different substituents (CH2OH, OH, and
OCH3) commonly observed in native lignin on the pyrolysis chemistry, four other compounds
with the same base structure of PPE were pyrolyzed. For exploring pyrolysis chemistry of α-O-4
compounds, benzylphenyl ether (BPE: PhCH2OPh) and BPE with methyl group at Cα were
pyrolyzed. Based on the pyrolysis results of the simplest model compounds (PPE and BPE) as
well as literature study on similar model compounds, we herein propose two initial bond
breakages: concerted pericyclic fragmentation for β-O-4 compounds, and homolytic cleavage for
α-O-4 compounds. The reaction network was further tested with more substituted model
compounds including four β-O-4 typed and one α-O-4 typed model compounds. Furthermore, we
attempted to describe pyrolysis behaviors of a trimer model compound having β-O-4 and α-O-4
linkages, which is more structurally similar to native lignin.
Experimental
Materials
The five β-O-4 typed, two α-O-4 typed, and a trimer lignin model compounds were
synthesized in NREL, and used in the current study (Figure 2). The β-O-4 typed model
compound includes 2-phenoxyphenylethanol (PPE), 2-phenoxyphenyl-1,3-propanediol (PPPD),
1-(4-hydroxyphenyl)-2-phenoxypropane-1,3-diol (HH), 1-(4-hydroxy-3-methoxyphenyl)-2-(2methoxyphenoxy)propane-1,3-diol (GG), and 2-(2,6-dimethoxyphenoxy)-1-(4-hydroxy-3,5-
9
dimethoxyphenyl)propane-1,3-diol (SS). The α-O-4 model compounds include benzylphenyl
ether (BPE) and 1-(phenoxyethyl)benzene (PEB). Trimer model compound containing both β-O4 and α-O-4 linkages is ((1-phenylethane-1,2-diyl)bis(oxy))dibenzene.
Micropyrolyzer-GC-MS/FID experiments
In the study the lignin model compounds were pyrolyzed at 500 °C, under an inert
atmosphere, using a single shot pyrolyzer (model 2020iS, Frontier Laboratory, Japan). A
micropyrolyzer is directly connected to a GC-MS/FID and the vapor products produced in a
pyrolyzer were immediately swept by 100 ml/min of helium, a carrier gas, into a GC injector
maintained at 400 °C. Approximately 500 µg of model compound were weighed by Mettler
Toledo microbalance, loaded in a sample cup, and were inserted into a pyrolysis tube prior to
pyrolysis. A residual solid in the sample cup was measured after the reaction to calculate char
yield. Non-condensable gases, carbon monoxide and carbon dioxide, were quantified by a DeJaye gas analyzer (equipped with an IR detector) that is connected to a split line from GC.
A GC-MS/FID was used to characterize the volatile products; a mass spectrometer (MS)
for identification and a flame ionization detector (FID) for quantification. The column used for
the chromatographic separation of the resultant volatile products was a Zebron ZB-5HT
INFERNO coated with 5 % phenyl and 95 % dimethylpolysiloxane with dimension of (30 m
length × 0.25 mm ID × 0.25 µm film thickness). The GC injector had a split ratio of 100:1, and
the GC oven temperature program started with a 3 minute hold at 35 °C followed by a 7 °C/min
heating rate up to 430 °C, held for 4 minutes. The peak assignment to a chemical was made by a
mass spectrometer with a NIST mass spectra library and/or the literatures[26, 27], and most of
the assignment was confirmed by injecting a pure standard in a GC-FID. To quantify each of the
volatile compounds, four known concentrations of each pure component diluted with methanol
10
were injected in to the GC-FID. The peak areas of each component were then integrated and
used to produce a calibration curve. For quantification of commercially unavailable compounds,
an FID response factor was approximated using the relationship of calibrated compounds
between an FID response and a molecule weight ratio (MWtotal/MWcarbon)[28]. The compounds
that used the estimated FID factor include 4-hydroxyacetophenone, 4-hydroxybenzyl alcohol,
((1-phenylethane-1,2-diyl)bis(oxy))dibenzene, 2,4-bis(1-phenylethyl)phenol, 4-propylguaiacol,
and dehydration products of β-O-4 model compounds. Based on the product characterization, a
mole balance throughout reactants and products of each model compound was conducted to
study reaction network.
Results and discussion
Thermal degradations of PPE and BPE model compounds
β-ether linked PPE
As Jarvis et al. proposed concerted pericyclic reaction (specifically, retro-ene
fragmentation) for pyrolysis of 2-phenethyl phenyl ether (PE: PhCH2 CH2OPh) having β-O-4
linkage, pyrolysis of PPE (PhCHOHCH2OPh), 1, structurally similar to PE, was initially
assumed to undergo the same reaction pattern. Based on the assumption and the product
identification of PPE pyrolysis, possible reaction network was illustrated in Figure 3. According
to the reaction network, β-O-4 linkage is initially cleaved by a retro-ene reaction through cyclic
transition state, 2, to yield two intermediates 3 and 4 with a mole ratio of 1:1. The reaction
intermediates with π bond are highly reactive at pyrolysis temperature, and as a result the further
reactions are likely to occur to form final pyrolysis products (which were detected by GCMS/FID). For example, phenol, 9, could be a result of a keto-enol tautomerization of
intermediate 4, and the other three products including benzeneacetaldehyde, 8, acetophenone, 6,
11
and benzaldehyde, 5, could be formed by three competitive pathways of intermediate 3. It is
likely that benzeneacetaldehyde, 8, is a result of keto-enol tautomerization of intermediate 7 that
is formed by dehydration and hydroxylation of intermediate 3. Formations of acetophenone, 6,
and benzaldehyde, 5, could be due to keto-enol tautomerization and demethylation followed by
tautomerization of intermediate 3, respectively. Surprisingly, the proposed pericyclic reaction
network is successfully supported by a mole balance throughout the reactants and products.
According to the mole balance, during the PPE pyrolysis 10.4 mmol of PPE was thermally
converted into intermediate 3 and 4, which are responsible for formations of the four thermal
fragments, benzeneacetaldehyde (6.2 mmol), 8, acetophenone (4.4 mmol), 6, and benzaldehyde
(0.5 mmol), 5, and phenol (9.7 mmol), 9, respectively. The remainder of PPE is dehydrated to (2phenoxyvinyl)benzene (1.2 mmol), 10, completing a mole balance for PPE pyrolysis. The close
agreement between the reaction network and the mole relationship highly suggests a concerted
retro-ene fragmentation as an initial reaction step for PPE pyrolysis. Another possible reaction
for PPE is a homolysis scission of Cβ-O and is demonstrated in Figure 4. However, the bond
fission is unlikely due to a generation of unstable 2-phenylethyl radical, 11, by homolysis.
Owing to electron deficiency in radicals, radicals tend to be stabilized through resonance
(delocalization of electron), and the radical stability affects the ease of homolytic cleavage.
However, in the case of PPE the resulting radical does not have resonance, which is caused by
absence of π bond next to carbon atom bearing a single, unpaired electron. Therefore, it is
concluded that homolytic fission of Cβ-O in PPE is not energetically favored. In fact, the bond
dissociation energy (BDE) was reported relatively high, 68 kcal/mol[25], compared to that of αO-4 typed BPE compound, 47-56 kcal/mol[20], which is proposed to undergo homolytic
cleavage in the study.
12
α-ether linked BPE
Unlike PPE, pyrolysis of BPE, 13, predominantly generated phenolic dimers such as
bibenzyl, 21, 2-benzylphenol, 25, and 4-benzylphenol, 22. The observation implies that the BPE
pyrolysis could be controlled by different reaction pathway from that of PPE pyrolysis.
There have been two possible reaction pathways for BPE pyrolysis, namely a Claisen
rearrangement and a homolytic dissociation. Ekpenyong et al. performed a thermolysis of BPE,
and proposed concerted pericyclic reaction to account for the experimental observations[29].
However, the proposed pericyclic reaction network presented in Figure 5 does not support the
formations of the major compounds of BPE pyrolysis in the current study, such as bibenzyl, 21,
2-benylphenol, 25, and 4-benzylphenol, 22. Moreover, if the pericyclic reaction predominantly
occurred in the current study, 2-(2-methylphenyl)phenol, 18, would be observed in a significant
amount, along with having equal moles of phenol, 9, and toluene, 19, generated. This is
inconsistent with the experimental observation. Therefore, it is highly doubtful that the pericyclic
reaction significantly occurs during BPE pyrolysis. In contrast to the pericyclic reaction, a freeradical reaction pathway initiated by homolysis of Cα-O bond successfully described
experimental results of BPE pyrolysis. According to the free-radical reaction network
demonstrated in Figure 6[30, 31], the benzyl radical, 20, and the phenoxy radical, 12, from
homolytic cleavage of the Cα-O bond undergo hydrogen abstraction to yield toluene, 19, and
phenol, 9, respectively. Bibenzyl, 21, one of the major products, is formed by a dimerization of
benzyl radical, 20. 4-benzylphenol, 22, and 2-benzylphenol, 25, are formed by tautomerizations
of intermediates 23 and 24, which are generated from interactions between phenoxy, 12, and
benzyl, 20, radicals, respectively. Importantly, the reaction network is strongly supported by a
generation of relatively stable radicals. Unlike PPE, homolytic cleavage of Cα-O bond in BPE is
13
energetically favored due to resonance stabilization energies of resulting benzyl, 20, and
phenoxy, 12, radicals (Figure 7). For the radicals, electron delocalization is likely to occur due to
existence of π bond adjacent to single, unpaired electron. In fact, relatively low BDE (47-56
kcal/mol) of Cα-O bond in BPE was reported[20]. In addition to the radical stability, a complete
mole balance of BPE pyrolysis supports the free-radical reaction network. Based on product
speciation and reaction network in Figure 6, it is anticipated that approximately 49.1 mmol of
both phenoxy, 12, and benzyl, 20, radicals are generated by a homolysis of Cα-O bond, which is
close to an actual mole of the converted BPE, 53.4 mmol. In fact, BPE acted as a radical initiator
in the pyrolysis (at 350 °C) of 1,3-diphenylpropane, dibenzyl ether and phenethyl phenyl
ether[32, 33], and thus at typical pyrolysis temperature of 500 °C a homolytic cleavage for BPE
is highly anticipated. In conclusion, BPE is thermally degraded at 500 C by free-radical reaction
initiated by homolytic fission of Cα-O. There are a trace amount of several products not
accounted for in the pathway, benzene and benzaldehyde. This might be due to a homolytic
cleavage of the bond between carbon in arene and adjacent oxygen. Due to relatively higher
BDE of the bond as compared to Cα-O bond, the bond breakage is expected to occur in a minor
extent.
Estimated rate constants
According to Vinu et al., reactions in micropyrolyzer-GC-MS/FID system that is used in
this study are kinetically controlled, in which fast-forming reactions (having transition state with
lower energy) are favored[34]. This is because reactions are irreversible due to short vapor
residence time, created by an immediate sweeping of volatile products with high flow rate of
helium gas. Thus, in order to determine which initial reaction breakage among pericyclic and
homolytic cleavage occurs predominantly during pyrolysis, it is reasonable to compare rate
14
constants of the different reaction pathways. In this section, rate constant, k, of concerted
pericyclic (retro-ene and Claisen) and homolytic dissociation to describe pyrolysis of PPE, 1, and
BPE, 13, were calculated based on activation energy, Ea, and pre-exponential factor, A, and
compared in Table 1. Pre-exponential factors for the concerted reactions, retro-ene fragmentation
for PPE and Claisen rearrangement for BPE, are estimated by those of pyrolysis of 2-phenethyl
phenyl ether and allyl vinyl ether, respectively[18, 25, 35, 36]. For a homolytic bond scission,
BDE are used to approximate the activation energies[20, 25], and large pre-exponential factors
are assumed[25]. As shown in Table 1, a rate constant of a retro-ene fragmentation is almost 9
times greater than that of a homolysis in the case of PPE pyrolysis, while for BPE pyrolysis the
rate constant for homolysis is three orders of magnitude greater than that of Claisen
rearrangement. This suggests that at pyrolysis temperature a retro-ene fragmentation is dominant
over a homolysis for PPE pyrolysis, whereas a homolytic fission is favored over a Claisen
rearrangement for BPE pyrolysis. Hence, the reaction networks proposed for pyrolysis of PPE
and BPE in the previous section were strongly supported by reaction kinetic.
Thermal degradations of more substituted model compounds
More substituted model compounds
Functional group commonly present in native lignin was added at a time into the simplest
model compounds (PPE and BPE), and their pyrolysis behaviors were investigated to determine
whether the proposed reaction networks (for PPE and BPE) are still applied for the more
substituted compounds. As shown in Figure 2, the compounds include β-ether linked PPPD, 26,
HH, 30, GG, 40, and SS, 55, and α-ether linked PEB, 67. For the substituted β-O-4 compounds,
the same retro-ene fragmentation as PPE, 1, was proposed as initial bond breakage. In fact, with
addition of alkyl groups a rate of retro-ene fragmentation was reported to be independent or even
15
increases, by steric effects[37, 38]. In contrast, PEB, 67, methyl-substituted α-O-4 compound is
assumed to undergo homolysis during pyrolysis, based on the generations of relatively stable
radicals through resonance.
Substituted β-O-4 model compounds
PPPD compound
In PPPD, 26, hydroxymethyl group (CH2OH) is attached to Cβ of PPE, 1. The addition of
the substituent was intended to resemble a propyl chain of a phenylpropane monomer, a basic
unit of a native lignin. If a cleavage of Cβ-O bond in PPPD, 26, occurs in the same pericyclic
fashion as PPE, 1, a formation of intermediates 27 and 4 (Figure 8) is expected. Under high
temperature, the resulting intermediates are subject to further reactions to generate phenolic
compounds (detected by GC-FID). Relating the intermediates to the final products provides the
following reactions we expected to occur: intermediate 4 is converted into more stable phenol, 9,
by keto-enol tautomerization, and intermediate 27 undergoes reactions including
tautomerization, dehydration, hydrogenation, deacetylation, deformylation, and demethylation,
to form benzaldehyde (82 mmol), 5, acetophenone (33 mmol), 6, benzeneacetaldehyde (31
mmol), 8, and cinnamyl alcohol (3 mmol), 29. It is noticed that first three products of PPPD, 26,
pyrolysis are also detected in PPE pyrolysis, due to the same β-O-4 structure of the two
compounds. Also, PPPD is dehydrated to generate 2-phenoxy-3-phenylprop-2-en-1-ol, 28.
Importantly, this pericyclic reaction network is successfully supported by a balanced mole
relationship throughout reactants and products. According to the mole balance, approximately
402 mmol of PPPD, 26, was thermally converted into the two intermediates 27 and 4 with a mole
ratio of approximately 1:1, which are responsible for five detected products, and the rest of
PPPD was dehydrated (210 mmol). Very little amount of char (0.5 wt%) was generated from
16
PPPD pyrolysis. A close consistency between the pericyclic reaction network and experimental
results strongly suggests a pericyclic reaction as main reaction for PPPD pyrolysis.
HH compound
HH, 30, a para-hydroxy substituted PPPD, 26, was used in the study to simulate H unit of
native lignin. Similar to previous β-O-4 compounds, HH is anticipated to undergo retro-ene
fragmentation to generate intermediates, 31 and 4, and further reactions of the intermediates
(Figure 9). The reactions include keto-enol tautomerization of intermediate 4 to generates phenol
(105.9 mmol), 9, and thermal decompositions/transformations of intermediate 31 to form 4hydroxybenzaldehyde (23.0 mmol), 35, 4-hydroxyacetophenone (17.7 mmol), 38, p-cresol (15.8
mmol), 39, p-coumaryl alcohol (12.1 mmol), 32, 4-hydroxybenzenepropanal (10.6 mmol), 34, 4hydroxybenzyl alcohol (10.5 mmol), 37, and 3-(p-hydroxyphenyl)-1-propanol (4.3 mmol), 33.
Due to an H unit characteristics of HH, some of the detected products are hydroxy-substituted,
which are typically observed in native lignin pyrolysis, such as p-cresol, 39, and 4hydroxybenzaldehyde, 35.[26] Among the proposed reactions above, three pathways that was
proposed in PPPD, such as deacetylation followed by tautomerization, deformylation followed
by tautomerization, and dehydration followed by hydrogenation and tautomerization were also
proposed in HH pyrolysis, generating major products, such as 4-hydroxybenzaldehyde, 35, 4hydroxyacetophenone, 38, and 4-hydroxybenzenepropanal, 34, respectively. It is not surprising
that similar reaction pathways are proposed from two β-O-4 compounds, because of a formation
of the similar types of intermediates through retro-ene fragmentation. It is also expected that
formaldehyde (77.4 mmol) was formed by further thermal degradations of 4hydroxybenzaldehyde, 35, 4-hydroxybenzyl alcohol, 37, 4-hydroxyacetophenone, 38, and pcresol, 39. The pericyclic reaction network successfully accounts for the product distributions,
17
implying that pericyclic reaction dominantly takes place during HH pyrolysis. According to the
mole balance, 247.0 mmol of HH was thermally converted; of which approximately 100.0 mmol
of HH was transformed into the various products, and the remaining HH was used for formations
of dehydration product (49.0 mmol), 36, and char (85.0 mmol). Additionally, a complete mole
balance of formaldehyde confirms that the pericyclic reaction is a main reaction channel for HH
pyrolysis; a summation (67.0 mmol) of moles of the possibly formed formaldehyde through the
thermal degradation is close to moles (77.4 mmol) of detected formaldehyde.
GG compound
Guaiacyl (G) unit predominantly present in softwood lignin was resembled by GG, 40, in
the study, with a methoxy group on ortho position of para-hydroxy substituted arene. On the
assumption of the same pericyclic reaction to occur as previous β-O-4 compounds, two
intermediates 41 and 42 are likely formed and the reactive intermediates are further subject to
similar types of reactions (to PPE, PPPD and HH) at pyrolysis temperature (Figure 10). For
example, keto-enol tautomerization of intermediate 42 forms guaiacol, 43, and for intermediate
41 various organic reactions such as dehydration, hydrogenation, tautomerization, deacetylation,
demethylation occurs to form coniferyl aldehyde, 46, coniferyl alcohol, 45, 4-(oxy-allyl)guaiacol, 47, 4-propyl guaiacol, 48, isoeugenol, 50, vanillin, 49, 4’-hydroxy-3’methoxyacetophenone, 54, 3-methoxybenzaldehyde, 52, acetaldehyde, and formaldehyde. Like
pyrolysis of PPPD, 26, and HH, 30, GG, 40 is also dehydrated and undergoes char-forming
reaction. Product distributions (by GC-MS/FID) are successfully explained by the above
pericyclic reaction network, which implies pericyclic reaction to occur significantly in GG
pyrolysis. According to the proposed reaction network and product distributions, during
pyrolysis approximately 120.0 mmol of GG, 40, undergo a cyclic transition state to generate the
18
intermediates which are responsible for the final products that are measured by GC-MS/FID. The
rest of GG (84 mmol) is involved in dehydrations and char-forming reactions, completing the
mole balance of GG pyrolysis. In short, a close consistency between the proposed pericyclic
reaction network and experimental observations strongly suggests pericyclic reaction pathways
for GG pyrolysis. It is noteworthy that guaiacyl-typed compounds (common products of native
lignin pyrolysis) were mainly observed in GG pyrolysis, such as coniferyl aldehyde, 46,
coniferyl alcohol, 45, 4-(oxy-allyl)-guaiacol, 47, 4-propyl guaiacol, 48, isoeugenol, 50, and
vanillin, 49 [26, 27].
SS compound
In the study, sinapyl (S) unit was resembled by SS, 55, and its pyrolysis behavior was
studied. Like the other β-O-4 compounds, experimental observation was well described by the
same retro-ene fragmentation followed by subsequent reactions of resulting intermediates
(Figure 11). Intermediate 57 is tautomerized to more stable syringol, 58, and intermediate 56 is
converted through various reactions into the rest of detected products, such as 4-hydroxy-3,5dimethoxyacetophenone, 64, 3,5-dimethoxy-4-hydroxycinnamaldehyde, 65, and sinapyl alcohol,
66. Additionally, the mole balance throughout the reactant and products suggests the pericyclic
reaction network as main reaction channel for SS pyrolysis. Of 188.5 mmol of SS that was
thermally converted, 148.5 mmol of SS was participated in retro-ene reaction to generate two
intermediates, 56 and 57, that are responsible for pyrolysis products. The rest of converted SS
was transformed into dehydration products such as 4-(2-(2,6-dimethoxyphenoxy)-3hydroxyprop-1-en-1-yl)-2,6-dimethoxyphenol, 63, and char.
19
Conclusion on pyrolysis of β-O-4 model compounds
Due to the absence of resonance structure for resulting radicals by homolysis, free-radical
reaction unlikely occurs during pyrolysis of β-O-4 compounds. Instead, retro-ene fragmentation
likely takes place as an initial bond cleavage to generate short-lived, reactive intermediates that
subsequently undergo thermal fragmentations. In fact, the rate constant calculation implies that
retro-ene fragmentation reaction was kinetically favored over homolysis, in the case of PPE. The
further reactions include tautomerization for intermediates 4, 42, and 57 to generate phenol (for
PPE, PPPD and HH), guaiacol (for GG) and syringol (for SS), respectively, and for counterpart
intermediates 3, 27, 31, 41, and 56 various reactions such as dehydration, hydrogenation,
deacetylation, deformylation and demethylation to generate remaining products of each model
compound. It is noted that throughout all β-O-4 compounds intermediates generated from
concerted channel undergo the fundamentally similar reaction pattern to generate products
(detected by GC-MS/FID), implying the proposed pericyclic reaction network as governing
reaction pathways during pyrolysis.
α-O-4 model compounds
PEB compound
Like BPE pyrolysis in Figure 7, homolysis of Cα-O bond in PEB, 67, is energetically
favored due to resonance structure of resulting radicals. Taking account of detected products
along with the radicals, it is proposed that the radicals undergo similar reactions to BPE, such as
hydrogen abstraction and oligomerization. According to the free-radical reaction network shown
in Figure 12, hydrogen abstraction of ethylbenzene, 68, and phenoxy, 12, radicals proceeds,
which generates ethylbenzene, 70,(some of which is subsequently dehydrogenated to styrene, 69)
and phenol, 9, respectively. The both radicals are also oligomerized into dimers such as 2-(1-
20
phenylethyl)phenol, 71, and 4-(1-phenylethyl)phenol, 72, and trimers such as 2,4-bis(1phenylethyl)phenol, 73. The measurement of dimers and trimers indicates a free-radical reaction
network as governing pathway, since radical can easily undergo oligomerization [39]. Although
a structure of the trimer molecule was not confirmed by pure standard, the structure is highly
likely based on mass fragmentation patterns (with matching score of over 900). Existence of
several peaks displaying the similar mass spectra indicates the presence of structural isomers of
the trimer. A mole balance of PEB pyrolysis is in accord with the free-radical reaction network.
Based on moles of major products, 431.2 mmol of ethylbenzene radical, 68, and 388.1 mmol of
phenoxy radical, 12, was formed, and the moles of radicals are close to a mole of converted PEB
(445.5 mmol). The agreement is a supporting evidence of a free-radical reaction network for
PEB pyrolysis.
Conclusion on pyrolysis of α-O-4 model compounds
Unlike β-O-4 compounds, α-O-4 model compounds (BPE and PEB) generate relatively
stable radicals through resonance structure. The radicals subsequently undergo hydrogen
abstraction, dimerization, and oligomerization, to yield final products. More importantly,
experimental observation is successfully accounted by free-radical reaction network (initiated by
homolysis). Additionally, homolytic cleavage of Cα-O bond in BPE is successfully supported by
rate constant calculation where homolysis is favored over pericyclic reaction.
Trimer model compound
In the previous sections, concerted pericyclic and free-radical reaction pathways have
been suggested for β-O-4 and α-O-4 linked compounds experimentally and computationally,
respectively. However, it is still unclear how the reaction pathways are applied to pyrolysis of
native lignin, which contains various linkages such as β-O-4, α-O-4, 5-5, and β-5. Thus, in the
21
study trimeric lignin model compound, 74, having the most common linkages, β-O-4 and α-O-4,
was tested.
Unfortunately, a marked amount of starting materials of trimer synthesis, such as PPE, 1,
and phenyl isobutyrate, was detected in the model compound, by Py-GC-MS run. As a result, a
complete product characterization was not performed for the trimer case; instead, product
identifications with relative intensity of each peak on the Py-GC chromatogram were used to
investigate a trimer reaction network.
Cα-O bond in α-O-4 compound (BDE 52 kcal/mol) is reported to be weaker than Cβ-O
bond in β-O-4 compound (BDE 65 kcal/mol). Besides, as demonstrated in Table 1 homolytic
cleavage of α-O-4 linkage occurs three orders of magnitude faster than pericyclic fragmentation
of β-O-4 linkage. Taking them into account, it is highly likely that homolytic cleavage of Cα-O
occurs faster than pericyclic fragmentation of Cβ-O during pyrolysis, thus leading to two
relatively stable radicals containing electron delocalization, 75 and 12 (Figure 13). Like α-O-4
compounds (BPE and PEB), it is highly expected that each radical undergoes a hydrogen
abstraction to generate 2-phenethyl phenyl ether (PE), 77, and phenol, 9. The resulting PE having
β-O-4 linkage, 77, could undergo the same pericyclic fragmentation as other β-O-4 compounds
(PPE, PPPD, HH, GG, and SS), to generate phenol, 9, and styrene, 69. If there was predominant
free-radical reaction for decomposition of PE, 77, products formed by hydrogen abstraction and
dimerization, such as ethylbenzene would be greatly observed. Another radical from the
homolysis of trimer, phenoxy radical, 12, is also expected to be dimerized, on the basis of an
observation of 2-phenoxyphenol, 76. A trimeric molecule of 2-(2-phenoxy-1phenylethyl)phenol, 78, corresponding to the largest peak on the Py-GC-MS/FID chromatogram,
was predicted based on its mass spectra (m/z: 196, 119, 197, 77, 104, 91, 103, 65, 183, 51);
22
however, the assignment was not confirmed by a pure standard due to its commercial
unavailability. The product is likely formed from an oligomerization of two initial radicals, 75
and 12, generated from Cα-O homolysis. It is noted that the dimers, 22 and 25, and trimer, 73, are
anticipated to be formed through the similar oligomerization of resulting radicals in pyrolysis of
α-O-4 compounds such as BPE and PEB, respectively. From the pyrolysis behaviors of trimer, it
appears that trimer pyrolysis is well explained by a combination of pericyclic and free-radical
reactions. Thus, it is believed that the proposed reaction channels can provide mechanistic
insights into pyrolysis of native lignin.
Conclusion
By performing a mole balance as well as a rate constant calculation for pyrolysis of each
model compound, we demonstrated that retro-ene fragmentation and homolysis predominantly
occur in pyrolysis of PPE and BPE, respectively. Moreover, the pathways for simplest model
compounds were successfully applied to more complicated compounds (PPPD, HH, GG, SS, and
PEB) and even trimer molecule. It is postulated that the difference in reaction network between
β-O-4 and α-O-4 compounds is caused by different stability of resulting radical (by resonance
structure) from homolysis. While homolysis of α-O-4 compound generates relatively stable
radicals by electron delocalization, that of β-O-4 compound generates relatively unstable radical
due to absence of resonance, thus making homolysis of β-O-4 compound unlikely. We believe
that the suggested reaction pathways for the model compounds will provide a mechanistic insight
to pyrolysis chemistry of native lignin.
References
1.
Lee, J., Biological conversion of lignocellulosic biomass to ethanol. Journal of
Biotechnology, 1997. 56(1): p. 1-24.
2.
Claassen, P., et al., Utilisation of biomass for the supply of energy carriers. Applied
microbiology and biotechnology, 1999. 52(6): p. 741-755.
23
3.
Zhang, S., et al., Process Modeling and Integration of Fuel Ethanol Production from
Lignocellulosic Biomass Based on Double Acid Hydrolysis. Energy & Fuels, 2009. 23(3):
p. 1759-1765.
4.
Kamm, B. and M. Kamm, Principles of biorefineries. Applied Microbiology and
Biotechnology, 2004. 64(2): p. 137-145.
5.
Kleinert, M. and T. Barth, Towards a Lignincellulosic Biorefinery: Direct One-Step
Conversion of Lignin to Hydrogen-Enriched Biofuel. Energy & Fuels, 2008. 22(2): p.
1371-1379.
6.
Stevens, C. and R.C. Brown, Thermochemical processing of biomass: conversion into
fuels, chemicals and power. Vol. 12. 2011: John Wiley & Sons.
7.
Huber, G.W., S. Iborra, and A. Corma, Synthesis of transportation fuels from biomass:
chemistry, catalysts, and engineering. Chemical reviews, 2006. 106(9): p. 4044-4098.
8.
Mohan, D., C.U. Pittman, and P.H. Steele, Pyrolysis of Wood/Biomass for Bio-oil: A
Critical Review. Energy & Fuels, 2006. 20(3): p. 848-889.
9.
Pandey, M.P. and C.S. Kim, Lignin depolymerization and conversion: a review of
thermochemical methods. Chem Eng Technol, 2011. 34(1): p. 29-41.
10.
Dolgonosov, B.M. and T.N. Gubernatorova, Modeling the biodegradation of
multicomponent organic matter in an aquatic environment: 2. Analysis of the structural
organization of lignin. Water Resources, 2010. 37(3): p. 320-331.
11.
Faix, O., D. Meier, and I. Grobe, Studies on isolated lignins and lignins in woody
materials by pyrolysis-gas chromatography-mass spectrometry and off-line pyrolysis-gas
chromatography with flame ionization detection. Journal of Analytical and Applied
Pyrolysis, 1987. 11(0): p. 403-416.
12.
Monteil-Rivera, F., et al., Isolation and characterization of herbaceous lignins for
applications in biomaterials. Industrial Crops and Products, 2013. 41(0): p. 356-364.
13.
Adler, E., Lignin chemistry—past, present and future. Wood Science and Technology,
1977. 11(3): p. 169-218.
14.
Holtman, K.M., An Investigation of the Milled Wood Lignin Isolation Procedure by
Solution-and Solid-State NMR Spectroscopy. 2004.
15.
Evans, R.J., T.A. Milne, and M.N. Soltys, Direct mass-spectrometric studies of the
pyrolysis of carbonaceous fuels: III. Primary pyrolysis of lignin. Journal of Analytical
and Applied Pyrolysis, 1986. 9(3): p. 207-236.
24
16.
Zakzeski, J., et al., The catalytic valorization of lignin for the production of renewable
chemicals. Chemical Reviews, 2010. 110(6): p. 3552-3599.
17.
Derkacheva, O. and D. Sukhov. Investigation of lignins by FTIR spectroscopy. in
Macromolecular Symposia. 2008. Wiley Online Library.
18.
Britt, P.F., A.I. Buchanan, and E.A. Malcolm, Thermolysis of phenethyl phenyl ether: a
model for ether linkages in lignin and low rank coal. The Journal of Organic Chemistry,
1995. 60(20): p. 6523-6536.
19.
Britt, P.F., et al., Flash Vacuum Pyrolysis of Methoxy-Substituted Lignin Model
Compounds. The Journal of Organic Chemistry, 2000. 65(5): p. 1376-1389.
20.
Schlosberg, R.H., W.H. Davis Jr, and T.R. Ashe, Pyrolysis studies of organic oxygenates.
2. Benzyl phenyl ether pyrolysis under batch autoclave conditions. Fuel, 1981. 60(3): p.
201-204.
21.
Schlosberg, R.H., et al., Pyrolysis of benzyl ether under hydrogen starvation conditions.
Fuel, 1981. 60(2): p. 155-157.
22.
Chu, S., A.V. Subrahmanyam, and G.W. Huber, The pyrolysis chemistry of a [small
beta]-O-4 type oligomeric lignin model compound. Green Chemistry, 2013. 15(1): p.
125-136.
23.
Smith, M.B., March's advanced organic chemistry: reactions, mechanisms, and
structure2013: Wiley. com.
24.
Klein, M.T. and P.S. Virk, Model pathways in lignin thermolysis. 1. Phenethyl phenyl
ether. Industrial & Engineering Chemistry Fundamentals, 1983. 22(1): p. 35-45.
25.
Jarvis, M.W., et al., Direct Detection of Products from the Pyrolysis of 2-Phenethyl
Phenyl Ether. The Journal of Physical Chemistry A, 2011. 115(4): p. 428-438.
26.
Faix, O., D. Meier, and I. Fortmann, Thermal degradation products of wood. Gas
chromatographic separation and mass spectrometric characterization of monomeric
lignin-derived products. Holz als Roh-und Werkstoff, 1990. 48(7-8): p. 281-285.
27.
Patwardhan, P.R., R.C. Brown, and B.H. Shanks, Understanding the fast pyrolysis of
lignin. ChemSusChem, 2011. 4(11): p. 1629-1636.
28.
Huang, Y., Q. Ou, and W. Yu, Characteristics of flame ionization detection for the
quantitative analysis of complex organic mixtures. Analytical Chemistry, 1990. 62(18): p.
2063-2064.
29.
Ekpenyong, I.A. and P.S. Virk, Pathways in coal thermolysis: a theoretical and
experimental study with model compounds. 1982.
25
30.
Ozawa, S., K. Sasaki, and Y. Ogino, Reaction of benzyl phenyl ether over various molten
metal catalysts. Fuel, 1986. 65(5): p. 707-710.
31.
Ozawa, S., T. Suenaga, and Y. Ogino, Effects of molten tin surface upon the reaction of
benzyl phenyl ether. Fuel, 1985. 64(5): p. 712-714.
32.
Gilbert, K. and J. Gajewski, Coal liquefaction model studies: free radical chain
decomposition of diphenylpropane, dibenzyl ether, and phenethyl phenyl ether via. beta.scission reactions. The Journal of Organic Chemistry, 1982. 47(25): p. 4899-4902.
33.
Gilbert, K.E., Coal liquefaction model studies: radical-initiated and phenol-inhibited
decomposition of 1, 3-diphenylpropane, dibenzyl ether and phenethyl phenyl ether. The
Journal of Organic Chemistry, 1984. 49(1): p. 6-10.
34.
Vinu, R. and L.J. Broadbelt, A mechanistic model of fast pyrolysis of glucose-based
carbohydrates to predict bio-oil composition. Energy & Environmental Science, 2012.
5(12): p. 9808-9826.
35.
Burrows, C.J. and B.K. Carpenter, Substituent effects on the aliphatic Claisen
rearrangement. 1. Synthesis and rearrangement of cyano-substituted allyl vinyl ethers.
Journal of the American Chemical Society, 1981. 103(23): p. 6983-6984.
36.
Schuler, F.W. and G.W. Murphy, The Kinetics of the Rearrangement of Vinyl Allyl
Ether1. Journal of the American Chemical Society, 1950. 72(7): p. 3155-3159.
37.
Houminer, Y., et al., Steric effects on pyrolysis reactions. Thermal retro-ene reaction of
pyrazineethanols. The Journal of Organic Chemistry, 1987. 52(18): p. 3971-3974.
38.
Paderes, G.D. and W.L. Jorgensen, Computer-assisted mechanistic evaluation of organic
reactions. 20. Ene and retro-ene chemistry. The Journal of Organic Chemistry, 1992.
57(6): p. 1904-1916.
39.
Togo, H., Advanced free radical reactions for organic synthesis2004: Access Online via
Elsevier.
26
Tables and figures
Figure 1. Softwood lignin structure [13]
27
(a)
(f)
(b)
(g)
(c)
(h)
(d)
(e)
Figure 2. (a) 2-phenoxyphenylethanol (PPE), (b) 2-phenoxyphenyl-1,3-propanediol (PPPD), (c)
1-(4-hydroxyphenyl)-2-phenoxypropane-1,3-diol (HH), (d) 1-(4-hydroxy-3-methoxyphenyl)-2(2-methoxyphenoxy)propane-1,3-diol (GG), (e) 2-(2,6-dimethoxyphenoxy)-1-(4-hydroxy-3,5dimethoxyphenyl)propane-1,3-diol (SS), (f) benzylphenyl ether (BPE), (g) 1(phenoxyethyl)benzene (PEB), and (h) ((1-phenylethane-1,2-diyl)bis(oxy))dibenzene.
28
Retroα
β
1
11.8
mmol
ene
β
α
+
2
3
11.1
4
9.7
mmol
mmol
Hydroxylation
Dehydration
Tautomerizatio
n
28
10
1.2 mmol
5
0.5 mmol
6
4.4 mmol
7
Tautomerizati
on
8
6.2 mmol
Figure 3. Pericyclic reaction network for PPE pyrolysis
9
9.7 mmol
29
1
High bond dissociation energy
(BDE): 60 – 70 kcal/mol
Figure 4. Homolytic cleavage for PPE
11
No resonance structure
12
30
α
Claisen
α
13
14
15
Tautomerization
17
Retro-ene
+2H
19
18
Figure 5. Pericyclic reaction network for BPE pyrolysis
18
30
9
16
31
9
31.2 mmol
19
1.3 mmol
H abstraction
Homolytic
cleavage
12
20
31
13
53.4 mmol
Tautomerization
22
6.6 mmol
21
16.0 mmol
Tautomerization
23
Figure 6. Free-radical reaction network for BPE pyrolysis
24
25
10.2 mmol
32
20
12
Low BDE: 47 – 57 kcal/mol
12
Resonance structures exist
Figure 7. Homolytic cleavage for BPE
Table 1. Estimation of rate constants at 500 °C
PPE
Ea, kcal/mol
A
k, s-1
Pericyclic
59.1
8E+12
1E-4
BPE
Homolysis
70.8
2E+15
2E-5
Pericyclic
54.9
5E+11
1E-4
Homolysis
54.7
1E+15
3E-1
Retro-ene
26
402 mmol
Dehydration
Tautomerization
27
149 mmol
4
198 mmol
9
198 mmol
Deformylation
Hydroxylation Deacetylation
Dehydration
Tautomerization
Tautomerization
33
28
210 mmol
8
31 mmol
Figure 8. Pericyclic reaction network for PPPD pyrolysis
5
82 mmol
29
3 mmol
6
33 mmol
32
12.1 mmol
33
4.3 mmol
34
10.6 mmol
35
23.0 mmol
Retroene
Tautomerizati
on
9
106.0 mmol
4
106.0 mmol
Dehydration
Cha
r
85.0 mmol
36
49.0 mmol
Figure 9. Pericyclic reaction network for HH pyrolysis
37
10.5 mmol
38
17.7 mmol
39
15.8 mmol
34
31
94.0 mmol
30
247.0 mmol
44
1.6 mmol
45
18.5 mmol
46
36.2
mmol
47
22.1
mmol
49
16.7
mmol
42
122.0 mmol
50
1.2 mmol
35
41
124.0 mmol
40
197.0 mmol
48
3.1 mmol
43
122.0 mmol
Char
51
69.0
mmol
15.0 mmol
52
6.7 mmol
Figure 10. Pericyclic reaction network for GG pyrolysis
53
0.3 mmol
35
2.0 mmol
54
15.2
mmol
39
0.8 mmol
36
59
7.6 mmol
60
5.1 mmol
61
12.9 mmol
62
3.0 mmol
36
55
188.5 mmol
56
143.9 mmol
57
153.6 mmol
58
153.6 mmol
Char
25.5 mmol
63
16.1 mmol
64
65
66
20.7 mmol
20.1 mmol
74.5 mmol
Figure 11. Pericyclic reaction network for SS pyrolysis
61.1 mmol
0.8 mmol
161.0 mmol
H abstraction
H abstraction
62.9 mmol
H abstraction
37
0.4 mmol
431.2 mmol
103.1 mmol
Figure 12. Free radical reaction network for PEB pyrolysis
Retro-ene
H abstraction
Oligomerization
H abstraction
Figure 13. Proposed reaction network for trimer pyrolysis
38
Dimerization
39
CHAPTER 3
DETAILED CHARACTERIZATION OF RED OAK-DERIVED PYROLYSIS OIL;
INTEGRATED USE OF GC, HPLC, IC, GPC AND KARL-FISCHER
A paper published in the Journal of Analytical and Applied Pyrolysis
Yong S. Choi, Patrick Johnston, Robert Brown, Brent Shanks, and Kyong Hwan Lee
Abstract
Red oak bio-oil obtained from fast pyrolysis in a fluidized bed reactor was fully analyzed using
gas chromatography/mass spectrometry (GC/MS), high-performance liquid chromatography
(HPLC), ion chromatography (IC), gel permeation chromatography (GPC) and Karl-Fischer
titration. Based on the chemical speciation, we achieved a high mass balance closure of bio-oil
(92 wt%). Of the analytical techniques, GC accounted for the largest portion of the bio-oil (25.7
wt%), followed by HPLC (15.4 wt%) and IC (8.1 wt%). Monosaccharides and disaccharides (not
detected by GC) were characterized by HPLC, and total sugars present in bio-oil were estimated
by running acid-hydrolysable sugars in the HPLC. Quantity of pyrolytic lignin, oligomer of
lignin-derived phenolic compounds, was determined as a GC-undetected portion of waterinsolubles of a whole bio-oil. Small standard deviation of each compound between duplicate runs
indicated a reproducibility of pyrolysis experiments and bio-oil characterizations validating the
detailed analytical approach used.
Introduction
Interest in liquid biofuels has grown due to the depletion of fossil resources, increasing
energy demand, and concerns over global warming. Although ethanol, produced by a
biochemical conversion of sugarcane and grain, has been commercially available, there has been
a debate over “food vs. fuel” owing to its usage of edible crop feedstocks. Therefore, significant
incentive exists to use non-edible crops to produce biofuels. Advanced biofuels, manufactured
40
from non-edible cellulosic biomass, are receiving significant attention with fast pyrolysis being
considered as one of the more promising technologies for economically producing advanced
fuels. Fast pyrolysis is defined as rapid thermal degradation of biomass in the absence of oxygen,
in order to produce a bio-oil as the primary product along with char and gases [1]. High heating
rates, short vapor residence times and rapid cooling of vapors are generally required to maximize
bio-oil yield during fast pyrolysis [1, 2]. In general, bio-oil is a complex mixture of more than
300 oxygenated organic compounds [3, 4], and can be broadly classified into anhydrosugars,
phenolics, furan/pyran derivatives and C2-C4 low molecular weight compounds. Due to number
of organic compounds generated, bio-oil has the potential to be upgraded into transportation
fuels and commodity chemicals via catalytic conversion [1, 5-7].
To design and develop robust downstream process strategies for upgrading bio-oil it is
necessary to know its full chemical properties but given its complexity a detailed chemical
speciation of bio-oil has been elusive. While as many of the plethora of species in bio-oil are
reported in literature, typically only compounds of interest among the identified compounds have
been quantified [3, 8, 9]. Additionally, most of the chemical analyses of bio-oil were performed
using GC which is limited to detection of volatile compounds. For example, only 30-35 wt% of
the total bio-oil was determined by quantification of 30-40 compounds using GC [8, 9]. Di Blasi
et al. characterized bio-oil from low-temperature pyrolysis of wood using GC-MS, with an
identification of 90 species and a quantification of 40 species, addressing 43 wt% of the bio-oil
[10]. Mullen et al. used both GC and HPLC for chemical analyses of bio-oil derived from
switchgrass and alfalfa, which accounted for 9-18 wt% of the bio-oil by quantification of 24
compounds out of 56 identified compounds [3]. Due to these incomplete characterizations, biooil analysis studies in literature have not attempted to perform detailed overall mass balance
41
closure for the bio-oils. In the current study, an integrated characterization strategy was to
determine the detailed chemical compositions of bio-oil obtained from fast pyrolysis of red oak.
Included in the studies were analyses conducted using GC, HPLC, IC, elemental analysis, and
Karl Fischer titration.
Experimental
Duplicate runs of red oak fast pyrolysis were conducted in a fluidized bed reactor and the
collected bio-oil from the runs was fully analyzed using a variety of analytical instruments as
discussed below.
Materials
Red oak, ground and sieved to 250-500 µm particles, was used as the feedstock for the
pyrolysis experiments. As shown in Table 1, ultimate and proximate analysis of the feedstock
was performed. The fluidizing medium used in the reactor was silica sand with an average
particle size of 512 µm (provided by Badger Mining Corporation) as this size provided good
fluidization in the bed for the carrier gas flowrate (10 L/min).
Pyrolysis in the fluidized bed reactor
A bench-scale bubbling fluidized bed reactor with a feed rate of 100 g/hr, shown in
Figure 1, was used in this study. Nitrogen was used to fluidize the inert silica sand bed during
pyrolysis. The fluidized bed reactor system was primarily composed of a screw feeder, fluidizedbed reactor, and series of condensers. Sufficient amount of feedstock was placed in the hopper
prior to supply the entire run, with the feedstock introduced to the reactor through an injection
auger. The auger was calibrated to deliver 100 g/hr of feedstock to the reactor. The reactor was
made of from a standard 316 stainless steel pipe with a 38.1 mm inner diameter and height of
0.34 m. The plenum region was located at the bottom of the reactor, which was designed to
42
preheat the nitrogen carrier gas. The reactor and plenum were encased with a heater and
maintained at 500 °C through a process control loop. Solid particles generated during pyrolysis
were collected downstream of the reactor in a series of cyclones. Hot vapors exiting the reactor
and cyclones were subjected to a cold quench of liquid nitrogen in which the vapors were
quenched at 90 °C. Then, the cooled vapors were passed through an electrostatic precipitator
(ESP) to remove aerosols and heavy molecules. Following the ESP, the vapors were passed
through a heat exchanger maintained at -10 °C and collected in two stages. Non-condensable
gases leaving the condenser system were passed through a wet test meter to measure their total
volumetric flowrate. The concentrations of these gases were measured using an online microGC-TCD (Varian CP 4900) equipped with three columns for different gases: Varian® Molesieve
5A column for H2, N2, CH4, and CO, Varian® PoraPLOT Q column for CO2, C2H2, C2H4, and
C2H6, and Varian® Al2O3 column for C3H8. After the reaction, the bio-oil yield was calculated
by considering the sum of the bio-oil fractions. The total char amount included solid particles
collected in the cyclones as well as in the reactor (minus the silica). The pyrolysis runs were
duplicated to evaluate the reproducibility of the experimental data.
Bed fluidization
To ensure the bed fluidization conditions created a high heat transfer rate, values of
U/Umf and deq/D for the 10 L/min of carrier gas rate were set between 2 and 3 and below 70 %,
respectively [11], by appropriate selection of the sand particle size. U is the superficial velocity
at the given flowrate at 500 °C, Umf is the minimum velocity for fluidization, deq is a volumeequivalent diameter of a bubble, and D is the bed diameter. Table 2 shows the process
parameters used for the fluidized bed runs.
Separation of whole bio-oil into water-soluble and water-insoluble fractions
43
Immediately after a pyrolysis reaction run, the bio-oil fractions collected during the
continuous reaction were mixed for 12 h with this mixture denoted as the whole bio-oil. The biooil was then separated into a water-soluble fraction and water-insoluble fraction by adding
distilled water with mixing to the whole bio-oil at a ratio of 1.7 wt water/wt whole bio-oil. The
sample was exposed to vortex mixing for 40 min followed by centrifugation at 3500 rpm for 10
min at which point a phase separation of the water-solubles and water-insolubles occurred. The
water-soluble fraction was collected by decanting the liquid phase from the water-insoluble
fraction. The three sets of bio-oil samples (whole bio-oil, water-soluble fraction, and waterinsoluble fraction) were individually analyzed for use in the overall mass balance.
Bio-oil characterization
There is no single analytical technique that can be used to perform a complete analysis of
bio-oil, due to its complex mixture of organic compounds. Thus, a range of analytical methods
were employed in the current study to elucidate and quantify as many species in the bio-oil as
possible. The full analysis of the bio-oils was conducted using GC, HPLC, IC, elemental
analysis, and Karl Fischer titration. GC was used for volatile compounds; HPLC for semi- or
non-volatile saccharides; IC for thermally labile organic acids; and Karl Fischer for water
content. To minimize aging effects during the characterization period, all bio-oil samples were
stored in tightly sealed Nalgene® bottle at 4 °C and all of the chemical analyses of the samples
were conducted within two weeks of their production.
Gas chromatography
A GC-MS/FID was used to characterize the volatile compounds in the bio-oil with the
mass spectrometer (MS) for identification and the flame ionization detector (FID) for
quantification. The column used for chromatographic separation was a Zebron ZB-1701 coated
44
with 14 % cyanopropylphenyl and 86 % dimethylpolysiloxane with dimensions of (60 m × 0.25
mm ID × 0.25 µm film thickness). The GC oven was programmed to hold at 35 °C for 3 minutes,
ramp at 5 °C/min to 300 °C, and then hold for 4 minutes. The injector was maintained at 300 °C
and employed a split ratio of 30:1. The flow rate was 1 mL/min of the helium carrier gas. The
mass spectrometer was configured for electron impact ionization, with a source/interface
temperature of 280 °C. The mass-to-charge ratio (m/z) values of the compound fragment ions
were recorded for each species exiting the GC column. Full scan mass spectra were acquired
from a 35 to 650 m/z at a scan rate of 0.5 seconds per scan. The peak assignment to a compound
was conducted using a NIST mass spectra library search in conjunction with the literature [1216]. The assignments were confirmed by injection of a pure standard into the GC.
To quantify each of the identified compounds, four known concentrations of each pure
component, diluted with methanol and phenanthrene (internal standard), were injected into the
GC-FID. The peak areas of each component and the internal standard were then integrated and
the relative areas were used to produce a calibration curve. The calibration curves for the pure
compounds all had strong linear relationships (R2 ≥ 0.90) between the FID response and
concentration of the standard. For the calibrated cyclic compounds, a relationship (R2 = 0.85)
between the ratio of total molecular weight to carbon weight (MWtotal/MWcarbon) and the FID
relative response factor was established (Figure 2) [17]. This correlation was used to
approximate the FID relative response for the cyclic compounds, which were unavailable
commercially. The compounds, which used the estimated FID response factor included, 2hydroxy-3-oxobutanal, 2-methyl-2-cyclopenten-1-one, 2-hydroxy-2-cyclopenten-1-one, 3methyl-2-cylcopenten-1-one, 4-hydroxy-5,6-dihydro-(2H)-pyran-2-one, 4-ethyl-2,6dimthoxyphenol, 2,6-dimethoxy-4-vinylphenol, and 2,6-dimethoxy-4-(1-propenyl)phenol. Peak
45
assignment to the unavailable compound was determined by comparing fragmentation patterns
of the unknown peaks with those reported in the literature [12-16].
Three bio-oil samples, whole bio-oil, water-soluble fraction, and water-insoluble fraction,
were individually prepared at approximately 15 wt% in a solution of methanol/acetone and
phenanthrene. The diluted sample was filtered through a Whatman® 0.45 micron glass microfiber
filter with1 µL injected into the GC.
Quantification of pyrolytic lignin
Although bio-oil water extraction method is widely used to separate and quantify
pyrolytic lignin in bio-oil [18], the amount of pyrolytic lignin in this study was determined as an
undetected portion (by GC) of the water-insolubles. It is expected that the pyrolytic lignin
measurement in this study was more accurate than the method reported in the literature, since the
water-insolubles that were considered as pyrolytic lignin in the prior method may contain
monomeric phenols, sugars, and low molecular weight compounds (LMWs), which can be
detected by GC. The non-GC detectables in the water-insoluble fraction were used with the
appropriate ratio for the mass balance for the whole bio-oil.
High-performance liquid chromatography
Water-soluble sugars
A HPLC equipped with a refractive index (RI) detector was used to characterize the
non/semi-volatile sugars that can be dissolved in water such as levoglucosan, cellobiosan,
xylose, and maltosan. The columns used were two Bio-Rad Aminex HPX-87P with a guard
column. The column temperature was 75 °C with a flow rate of 0.6 mL/min 18.2Ω distilled
water. The RI detector was calibrated with the non-volatile sugars diluted into five
concentrations (0-10 mg/ml) with distilled water. Calibration standards were obtained from
46
Carbosynth (levoglucosan, cellobiosan and maltosan) and Thermo Fisher Scientific (xylose).
Approximately 0.5 g of a bio-oil sample was dissolved in 5 ml of distilled water and well mixed
with a vortex mixer for 20 min. The resulting solution was then filtered through a Whatman®
0.45 micron glass microfiber filter and 25 µL were injected into the HPLC.
Total sugars
Hydrolysable sugars in bio-oil were quantified by HPLC with a Shodex refractive index
detector to estimate total sugar amount in the bio-oil. Under dilute acid condition,
oligosaccharides were hydrolyzed into monosaccharides by the breaking of glycosidic bonds [19,
20], and the monosaccharides were further converted into glucose or sorbitol (for C6 sugars) and
xylose (for C5 sugars). Sugars in the bio-oil were detected in the form of glucose, sorbitol or
xylose by the HPLC system and the detected monosaccharide amount was used to approximate
the total sugars in the bio-oil. The column used was a HyperRez XP Carbohydrate (300 mm ×
7.7 mm 8 µm particle size). The mobile phase was 18.2Ω deionized water with a flow rate of 0.2
mL/min and the column was set at 55 °C. Approximately 60 mg of bio-oil and 6 mL of 400 mM
H2SO4 were added to a sealed glass vial, and the sugars were acid-hydrolyzed in an oil bath at
conditions of 125 °C for 60 min. The hydrolyzed solution was then filtered through a Whatman®
0.45 micron glass microfiber filter, and 25 µL were injected into the HPLC.
Ion chromatography
Carboxylic acids including acetic acid, formic acid, glycolic acid, and propanoic acid,
were characterized by IC as their broad peak shapes lead to inaccurate quantification by GC-FID
[9]. The IC system used was a Dionex ICS3000 equipped with a conductivity detector and an
Anion Micromembrane Suppressor AMMS-ICE 300. The suppressor regenerant was 5 mM
tetrabutylammonia hydroxide at a flow rate of 4-5 mL/min. The mobile phase was 1.0 mM
47
heptaflourobutyric acid used in an IonPac® ICE-AS1 analytical column at a flow rate of 0.120
mL/min at 19 °C. The conductivity detector was calibrated with a standard solution (purchased
from Inorganic Ventures) containing the four acids diluted into five concentrations (10-200
mg/L). Peak assignment was determined by comparing peak retention times of the standard to
those in the bio-oil sample. The bio-oil samples were prepared using 6 mL distilled water and 1.5
mL of methanol. The diluted samples were filtered with a syringe filter (0.45 µm) prior to
injection.
Elemental analysis
Ultimate analysis of the biomass and bio-oil samples was conducted using an Elementar®
elemental analyzer (vario MICRO cube). During elemental analysis, a sample was combusted at
900 °C, and the combustion products, carbon dioxide, water and nitric oxide, were characterized
by a TCD. The weight percentages of C, H and N were calculated based on the amount of the
combustion products. Oxygen was calculated by difference. Approximately 5 mg of sample was
weighed and inserted into the combustion chamber for the analysis.
Thermogravimetric analysis
Proximate analysis of the feedstock was performed in a TGA with the following method:
approximately 20 mg of biomass was heated in 100 mL/min of N2 from 25 °C to 105 °C at 10
°C/min and held at 105 °C for 40 min to eliminate moisture in the biomass. The temperature was
further increased to 900 °C at 10 °C/min and held for 20 min to quantify the volatiles. Then, 100
mL/min of air was introduced at 900 °C for 30 min to combust the remainder of the sample, with
the subsequent weight loss being considered fixed carbon. The amount of ash was determined by
the residual weight of the sample.
48
Karl Fischer titration
The water content of the bio-oil was determined by Karl Fischer titration (titrator).
Gel permeation chromatography
GPC is a type of size exclusion chromatography where the separation occurs on the basis
of the size of the analytes. Small analytes elute slower than large analytes, since small molecules
are retained longer in the pores of the stationary phase of the column. GPC analysis was
performed on a HPLC system (Dionex Ultimate 3000) equipped with a Shodex Refractive Index
(RI) and Diode Array Detector (DAD) detector. Two Agilent PLgel 3 µm 100Å, 300 mm × 7.5
mm columns and one Mesopore 300 mm × 7.5 mm analytical column were used. The mobile
phase (tetrahydrofuran, THF) flowed into the column at a rate of 1.0 mL/min at 25°C. Seven
polystyrene standards with peak molecular weights of 162, 380, 580, 970, 1930, 2900, and 3790
were used for molecular weight calibration. A 0.02 g bio-oil sample was dissolved in a 10 mL of
THF, with the solution filtered using a syringe filter (0.45 µm) prior to the GPC analysis. Data
were acquired and evaluated using Dionex Chromeleon software, Version 6.8.
Results and discussions
Biomass properties
According to composition analysis of red oak reported in Table 3, it is composed of 40.7
wt% cellulose, 22.8 wt% hemicellulose, and 33.3 wt% lignin, indicating that approximately twothirds of the feedstock is hollocellulose. Proximate and ultimate analyses of the as-received red
oak feedstock are given in Table 1. According to the proximate analysis, the feedstock had 8.3
wt% moisture, which is in the typical range. Based on the elemental analysis, higher heating
value (HHV, MJ/kg) of red oak was calculated by the following formula, HHV = {33.5[C] +
142.3[H] - 15.4[O] – 14.5[N]} x 10-2 [21]. Carbon, hydrogen, and nitrogen were determined in
49
the as-received feedstock and are shown in Table 4. The oxygen content was calculated by
difference. The elemental analysis of the bio-oil yielded an average chemical formula of
C2.2H2.3O, which corresponded to approximately 35 wt% oxygen (compared to the 47 wt%
oxygen as determined by ultimate analysis).
Biomass pyrolysis
A duplicate red oak pyrolysis runs at 500 °C was performed in the fluidized bed reactor
to generate bio-oil, char, and non-condensable gases. The overall mass balance closure (Table 5)
for the liquid, solid and gas were 86.2 % and 85.5 %, respectively, for the duplicate runs. The
unknown masses might be attributed to bio-oil residue left in pipes and fittings and there was
likely some error associated with measuring the non-condensable gases. Upon fast pyrolysis, the
red oak was transformed into liquid bio-oil (61.1 wt%), char (11.6 wt%) and gases (13.2 wt%).
In the bio-oil, a 26.8 wt% of water was present, which allowed the resulting bio-oil to flow.
Among the non-condensable gases generated, carbon dioxide and carbon monoxide were
predominant, supporting the importance of decarboxylation and decarbonylation for biomass
deoxygenation during pyrolysis [22]. Sufficient reproduction of the process operating conditions
(temperatures, feeding rate, vapor residence time, and condensing rate) between runs was
achieved such that the results were reproducible, which was indicated by the small standard
deviation between the lumped products of the duplicate runs.
Bio-oil characterization
The whole bio-oil collected during the continuous run was separated into water-soluble
and water-insoluble fractions by water extraction of the whole bio-oil. The resulting distribution
for the whole bio-oil as shown in Table 6 demonstrated that water-soluble fraction constituted
approximately 71 wt% of the whole bio-oil. The 29 wt% remainder, which was water-insoluble
50
fraction, was close to the lignin content (33.3 wt%) in the original feedstock. This result was
consistent with expectation as the bio-oil was the major product from fast pyrolysis and the
lignin-derived product in the bio-oil was primarily pyrolytic lignin which would be hydrophobic.
Table 7 lists 50 compounds and additional groups of compounds present in at least 0.01
wt% that were identified and quantified by the integrated analytical techniques. The values
shown are for the analysis of each of the water-soluble and water-insoluble fractions as well as
the whole bio-oil. Statistical information for the duplicate runs is given in the fourth and fifth
columns. Theoretical composition of whole bio-oil was calculated using yield of each of watersoluble and water-insoluble fractions and its distribution in the whole bio-oil (from Table 6), and
was demonstrated in sixth column. For the whole bio-oil, actual yields were compared to
theoretical values in the last column. The calculations were intended to show how precisely and
accurately the pyrolysis experiments and chemical analyses were performed in the study.
Although some sugars and pyrolytic lignin were not individually determined due to limitations of
the analytical techniques, their amounts was found in lumped groups, which were also included
in the overall mass balance.
Gas chromatography analysis of bio-oils
When the diluted bio-oil sample was injected into the GC system, only compounds
volatile at 300 °C were able to pass through a capillary column to be separated, and detected by
MS and FID. 40 major peaks were identified by the MS and quantified by the FID, contributing
to approximately 25.3 wt% of the whole bio-oil. There is a potential to further improve the mass
balance if comprehensive two-dimensional gas chromatography with its better separation of biooil species was used[23]. However, reasonable mass balances can be achieved even without
using that more extensive approach. Expectedly, carbohydrate-derived compounds were
51
predominant, constituting approximately 80 % of the GC-detectables. For example, the weight
percentages of glycolaldehyde, formaldehyde, methyl glyoxal, and acetol were 5.6, 3.5, 2.4, and
2.1 wt%, respectively, of the whole bio-oil. On the other hand, relatively small amounts of
lignin-derived compounds were detected. Guaiacyl (G)- and syringyl (S)-type phenols were
dominant over hydroxyl (H)-type phenol among the detected phenols, since hardwood lignin is
mainly composed of G and S units [24]. In fact, nearly 95 % of the GC-detectable phenols were
hydroxyphenols with methoxy groups attached.
The amount of the pyrolytic lignin measured was 16 wt%, which was lower than
typically reported in the literature (25 wt%) [1], which was determined by the water extraction
method of Meier et al [18]. The lower value reported in the current work resulted from the
method modification, which quantified some of the species in pyrolytic lignin thereby removing
them from the lumped group. The quantification process in this study used an extra step to filter
out water soluble compounds (phenol, sugars, and LMWs) still present in the water-insoluble
fraction prior to injection into the GC. While the quantity of the remaining pyrolytic lignin after
the extra extraction step was accurately obtained, its exact structure was still unknown. However,
it is widely accepted that pyrolytic lignin is comprised of oligomers resulting from lignin-derived
phenolic compounds, which was evidenced by GPC analysis as discussed below.
High performance liquid chromatography analysis of bio-oils
Although GC has been the most powerful and popular analytical technique for bio-oil
characterization, the technique is limited to detection of volatile compounds determined at a
given injector temperature. As a consequence, a significant amount of semi-/nonvolatile species
cannot be characterized, leading to a low bio-oil mass balance closure [8-10]. Additionally, the
existence of high molecular weight compounds was supported by GPC analysis as presented in
52
Table 8. For this study, a HPLC with two Bio-Rad Aminex HPX-87P columns was used to detect
high molecular weight compounds in bio-oil, such as levoglucosan, cellobiosan, and maltosan.
Whereas, the total sugar estimated amount in bio-oil was determined with a HPLC with a
HyperRez XP Carbohydrate after acid-hydrolysis of bio-oil samples.
In the HPLC analysis, identification of a sugar was performed by comparing its peak
retention time with that of each sugar in a standard mixture. Quantification of levoglucosan,
which has been characterized by GC in many research groups [8-10, 25], was instead determined
using HPLC in the study since its boiling point is similar to the injector temperature that was
employed for the GC analysis. The close alignment of temperature might lead to only partial
volatilization of the levoglucosan and thus underestimation of its content in the bio-oil using
typical GC methods. In fact, the measured levoglucosan yield increased approximately 30 %
when characterized by HPLC, compared to GC. Anhydrosugars derived from cellulose, 1,6anhydro-β-D-glucopyranose (levoglucosan), 1,6-anhydro-α -D-maltose (maltosan) and 1,6anhydro-β-D-cellobiose (cellobiosan) were detected as 5.0, 1.6, and 1.2 wt%, respectively.
Notably, the yield of maltosan having an α(1→4) glycosidic linkage was appreciable, although
cellulose is a polysaccharide consisting of glucose connected by β(1→4) glycosidic linkage.
Houminer et al claimed that formation of maltosan takes place by an inversion of C1 in
cellobiosan which is a result of a levoglucosan dimerization creating a β(1→4) glycosidic
linkage [26]. Yields of glucose, xylose, and sorbitol were summed to calculate total sugars in the
bio-oil. It was estimated that around 15 wt% of the bio-oil consisted of sugars, including watersoluble sugars. Although approximately two-thirds of the feedstock was carbohydrates, only a
small portion of the sugars were retained through fast pyrolysis, demonstrating a significant
thermal degradation of the sugars during pyrolysis. Although a half of the quantified sugars by
53
acid-hydrolysis were not structurally identified in the study, it was speculated that the
unidentified sugars were dimers of levoglucosan with glycosidic linkage of α or β(1→2) and α or
β(1→3), anhydrous dimers of xylose with all possible glycosidic linkages, or trimers/tetramers
of the hexose and pentose. Those sugars were labeled as unidentified oligosaccharides in Table
7.
Ion chromatography analysis of bio-oils
Carboxylic acids as reported in Table 7 were characterized by IC rather than GC, due to
their thermal instability and broad peak shape in the GC chromatograms. Whole bio-oil consisted
of approximately 7 wt% organic acids including acetic, propanoic, glycolic, and formic acids.
Along with glycolaldehyde, acetic acid was a major product of red oak pyrolysis, corresponding
to over 5 wt% yield.
Gel permeation chromatography analysis of bio-oils
As determined by GPC, the weight-average molecular weights (Mw) for the whole biooils generated from the duplicate runs were 396 and 388, while the number-average molecular
weights (Mn) were 215 and 214 (Table 8). The polydispersities (PD = Mw/Mn) for the bio-oils
were 1.84 and 1.81, indicating a wide range of molecular weight distributions. It was noted that
the weight-average molecular weight was higher than any identified/quantified compound in the
bio-oils, such as levoglucosan (162.1), cellobiosan (324.3), and 4-allyl-2,6-dimethoxyphenol
(194.2). This result suggested that the unidentified species (some sugars and pyrolytic lignin) in
the bio-oil were mainly comprised of heavy molecules (dimers and trimers).
Product distributions of whole bio-oil, water-solubles, and water-insolubles
As shown in Table 7, the integration of analysis techniques led to a mass balance closure
for the whole bio-oil of 92 wt%, with 27 wt% by Karl Fischer titration, 26 wt% by GC, 15 wt%
54
by HPLC, 7 wt% by IC and 16 wt% pyrolytic lignin. Of the analytical techniques employed, GC
as expected gave quantification for the largest organic portion of the bio-oil, but HPLC and IC,
which have not been normally used for bio-oil characterization, accounted for an appreciable
amount (22 wt%). With the concomitant quantification of the pyrolytic lignin, the remaining
mass could be at least partially due to the presence of small unidentified peaks in the GC
chromatograms. High mass closure as well as data reproducibility/accuracy as indicated by the
low standard deviations and the values of A/B close to unity (Table 7) demonstrated successful
and reliable pyrolysis experiments and chemical analyses of the collected bio-oil. As shown in
Figure 3, pyrolytic lignin (derived from lignin, 16 wt%), low molecular weight compounds,
LMWs (derived from carbohydrates, 16 wt%) and sugars (derived from carbohydrates, 15 wt%)
were the most predominant organic compounds in the whole bio-oil. Carboxylic acids, phenols,
furans, cyclics represented smaller fraction of the whole bio-oil, making up 7 wt%, 4 wt%, 2
wt% and 2 wt% of the whole bio-oil, respectively. It is noted that relatively large difference was
observed between runs for LMWs, resulting from broad peak shapes of glycolaldehyde and
formaldehyde on GC analysis.
Breakdowns of the water-soluble and water insoluble fractions are shown in Figures 4
and 5. The water-soluble fraction was about 70 wt% of whole bio-oil and mainly consisted of
carbohydrate-derived compounds such as sugars, LMWs, and acids. In the water-soluble
fraction, there were small amounts of hydroxyphenols with two methoxy groups attached, such
as 2,6-dimethoxyphenol and 4-methyl-2,6-dimethoxyphenol, derived from lignin. In contrast, a
majority of the water-insolubles were lignin-derived compounds, including pyrolytic lignin (55.6
wt%) and phenols (10.4 wt%). The phenols present in the water-insolubles were mainly
hydroxyphenols with one or two methoxy groups attached, such as 2,6-dimethoxy-4-(1-
55
propenyl) phenol, 2,6-dimethoxy-4-vinylphenol, isoeugenol, 2,6-dimethoxyphenol and 4-methyl2,6-dimethoxyphenol. This speciation was likely due to hardwood lignin being composed of
guaiacyl (with one methoxy group) and syringol (with two methoxy groups) units. Due to the
difficulty in achieving completely clean phase separation, a small amount of water-soluble
compounds such as sugars and acids was detected in the water-insoluble fraction.
Elemental balances throughout bio-oil, char, and gases
Elemental analysis of the whole bio-oil and char, and non-condensable gases product
distributions were used to perform carbon, hydrogen, and oxygen balances on the pyrolysis
products, which is shown in Table 9. Taking a basis of 100g of red oak feedstock pyrolyzed, the
overall balances on C, H, and O that could be accounted for in the balances from the pyrolysis
products were 82.2, 80.2, and 90.2 wt%, respectively, 14.1 g of the feedstock was note directly
unaccounted. Based on the elemental balances, the unaccounted mass would effectively have an
empirical formula of C2.3H4.6O. The unaccounted mass could be attributed to bio-oil and char
residues in the pipes and/or fittings which were not measured in the study. Of the C, H, and O,
about 52% of the feedstock carbon was recovered in the form of bio-oil, primarily due to char
production, whereas approximately 70 wt% of the hydrogen and oxygen in feedstock ended up in
the bio-oil.
Conclusions
Bio-oils obtained from fast pyrolysis of red oak in a fluidized bed reactor were
chemically analyzed using a broad range of analytical techniques. In contrast to the majority of
pyrolysis studies that tend to focus only on the quantification of several key compounds (acetic
acid, glycolaldehyde, and levoglucosan), 52 species were identified and quantified in the bio-oils
and thereby achieving a high mass balance closure on the bio-oil of nearly 92 wt%. Duplicate
56
pyrolysis runs demonstrated the high degree of data reproducibility. Although GC-detectables
occupied the largest portion of the organics in the bio-oil, the quantification of species by HPLC
and IC appreciably contributed to the total mass balance closure. These results demonstrate the
importance of using HPLC and IC to fully characterize bio-oil. The precise structure of all biooil products was not achieved as a significant portion (16 wt%) of the bio-oil was pyrolytic lignin
and half of the quantified sugars appeared to be present in the form of dimers or oligomers, based
on GPC analysis. While still not a complete compositional picture for the fast pyrolysis of
biomass, this study provided the most complete analysis of composition while introducing
complementary analytical techniques that refined the quantification. Understanding the true and
detailed chemical composition of bio-oil as determined by the integrated analytical techniques in
the study will be critical in identifying and solving problems that arise in optimizing downstream
processes for upgrading the bio-oil into fuels and chemicals.
References
1.
Mohan, D., C.U. Pittman, and P.H. Steele, Pyrolysis of Wood/Biomass for Bio-oil: A
Critical Review. Energy & Fuels, 2006. 20(3): p. 848-889.
2.
Stevens, C. and R.C. Brown, Thermochemical processing of biomass: conversion into
fuels, chemicals and power. Vol. 12. 2011: John Wiley & Sons.
3.
Mullen, C.A. and A.A. Boateng, Chemical Composition of Bio-oils Produced by Fast
Pyrolysis of Two Energy Crops†. Energy & fuels, 2008. 22(3): p. 2104-2109.
4.
Diebold, J.P., A review of the chemical and physical mechanisms of the storage stability
of fast pyrolysis bio-oils2000: National Renewable Energy Laboratory Golden, CO.
5.
Huber, G.W., S. Iborra, and A. Corma, Synthesis of transportation fuels from biomass:
chemistry, catalysts, and engineering. Chemical reviews, 2006. 106(9): p. 4044-4098.
6.
Huber, G.W., et al., Production of liquid alkanes by aqueous-phase processing of
biomass-derived carbohydrates. Science, 2005. 308: p. 1446-2079.
7.
Sarma, A.K., D. Konwer, and P. Bordoloi, A Comprehensive analysis of fuel properties
of biodiesel from Koroch seed oil. Energy & fuels, 2005. 19(2): p. 656-657.
57
8.
Ingram, L., et al., Pyrolysis of wood and bark in an auger reactor: physical properties
and chemical analysis of the produced bio-oils. Energy & Fuels, 2007. 22(1): p. 614-625.
9.
Pittman Jr, C.U., et al., Characterization of Bio-oils Produced from Fast Pyrolysis of
Corn Stalks in an Auger Reactor. Energy & Fuels, 2012. 26(6): p. 3816-3825.
10.
Branca, C., P. Giudicianni, and C. Di Blasi, GC/MS characterization of liquids generated
from low-temperature pyrolysis of wood. Industrial & engineering chemistry research,
2003. 42(14): p. 3190-3202.
11.
Kunii, D. and O. Levenspiel, Fluidization engineering. Vol. 2. 1991: ButterworthHeinemann Boston.
12.
Faix, O., et al., Thermal degradation products of wood: gas chromatographic separation
and mass spectrometric characterization of polysaccharide derived products. Holz als
Roh-und Werkstoff, 1991. 49(5): p. 213-219.
13.
Faix, O., D. Meier, and I. Fortmann, Thermal degradation products of wood. Gas
chromatographic separation and mass spectrometric characterization of monomeric
lignin-derived products. Holz als Roh-und Werkstoff, 1990. 48(7-8): p. 281-285.
14.
Patwardhan, P.R., R.C. Brown, and B.H. Shanks, Understanding the fast pyrolysis of
lignin. ChemSusChem, 2011. 4(11): p. 1629-1636.
15.
Patwardhan, P.R., et al., Product distribution from fast pyrolysis of glucose-based
carbohydrates. Journal of Analytical and Applied Pyrolysis, 2009. 86(2): p. 323-330.
16.
Patwardhan, P.R., R.C. Brown, and B.H. Shanks, Product distribution from the fast
pyrolysis of hemicellulose. ChemSusChem, 2011. 4(5): p. 636-643.
17.
Huang, Y., Q. Ou, and W. Yu, Characteristics of flame ionization detection for the
quantitative analysis of complex organic mixtures. Analytical Chemistry, 1990. 62(18): p.
2063-2064.
18.
Scholze, B. and D. Meier, Characterization of the water-insoluble fraction from pyrolysis
oil (pyrolytic lignin). Part I. PY–GC/MS, FTIR, and functional groups. Journal of
Analytical and Applied Pyrolysis, 2001. 60(1): p. 41-54.
19.
Helle, S., et al., A kinetic model for production of glucose by hydrolysis of levoglucosan
and cellobiosan from pyrolysis oil. Carbohydrate research, 2007. 342(16): p. 2365-2370.
20.
Lian, J., et al., Separation, hydrolysis and fermentation of pyrolytic sugars to produce
ethanol and lipids. Bioresource technology, 2010. 101(24): p. 9688-9699.
21.
Demirbaş, A., Calculation of higher heating values of biomass fuels. Fuel, 1997. 76(5): p.
431-434.
58
22.
Zabeti, M., et al., < i> In situ</i> catalytic pyrolysis of lignocellulose using alkalimodified amorphous silica alumina. Bioresource technology, 2012. 118: p. 374-381.
23.
Marsman, J., et al., Identification of components in fast pyrolysis oil and upgraded
products by comprehensive two-dimensional gas chromatography and flame ionisation
detection. Journal of Chromatography A, 2007. 1150(1): p. 21-27.
24.
Dolgonosov, B.M. and T.N. Gubernatorova, Modeling the biodegradation of
multicomponent organic matter in an aquatic environment: 2. Analysis of the structural
organization of lignin. Water Resources, 2010. 37(3): p. 320-331.
25.
Patwardhan, P.R., et al., Distinguishing primary and secondary reactions of cellulose
pyrolysis. Bioresource Technology, 2011. 102(8): p. 5265-5269.
26.
Houminer, Y. and S. Patai, Thermal polymerization of levoglucosan. Journal of Polymer
Science Part A‐1: Polymer Chemistry, 1969. 7(10): p. 3005-3014.
Tables and figures
Table 1. Ultimate and proximate analyses of the red oak feedstock
Ultimate analysis
Proximate analysis
wt%
wt%
Carbon
45.8
Moisture
8.3
Hydrogen
6.7
Volatiles
78.6
Nitrogen
0.1
Fixed carbon
12.7
Oxygen
47.4
Ash
0.3
Total
100
Total
99.9
HHV, MJ/kg
17.6
HHV = {33.5[C] + 142.3[H] - 15.4[O] – 14.5[N]} x 10-2
59
Table 2. Process parameters used for the fluidized bed pyrolysis
Residence time,
Sand size,
N2 flow rate,
seconds
µm
L/min
1.3
512
10.0
Where:
U- Superficial velocity at given N2 flowrate at 500 °C
Umf- Minimum fluidization velocity
deq- Volume-equivalent diameter of bubble
D- Bed diameter
U/Umf
deq/D
2.8
65.3
60
Table 3. Compositional analysis of the red oak feedstock
Component
Composition, wt%
Cellulose
40.7
Hemicellulose
22.8
Lignin
33.3
Ash
0.4
Total
97.2
Conducted by the protocol of the NREL Chemical Analysis and Testing Standard Procedures:
NREL LAP, TP-510-42618.
Table 4. Elemental analysis of the whole bio-oil
Run 1, %
Run 2, %
Average
STD
RSD
Carbon
57.5
61.3
59.4
1.9
3.1
Hydrogen
5.9
4.6
5.2
0.6
12.3
Oxygen
36.6
34.1
35.3
1.2
3.5
Numbers in the first two columns are in wt% based on dry feedstock. Oxygen was calculated by
difference.
61
Table 5. Overall mass balance for the pyrolysis runs
Run 1
Run 2
Average
STD
61.90
60.33
61.11
0.79
Organics
45.36
44.02
44.69
-
Water
16.54
16.31
16.42
0.11
Char
11.30
11.86
11.58
0.28
Non-condensable gases
13.04
13.34
13.19
0.15
H2
0.01
0.01
0.01
0.00
CH4
0.50
0.47
0.48
0.01
C2H6
0.04
0.02
0.03
0.01
C2H4
0.16
0.17
0.16
0.01
CO
5.39
5.29
5.34
0.05
CO2
6.94
7.38
7.16
0.22
Total
86.24
85.53
85.89
0.35
Bio-oil
All numbers in wt% based on feedstock
Table 6. Weight percentage for the water-soluble and water-insoluble fractions in the whole biooil
Run 1
Run 2
Average
STD
Water-solubles
69.9
71.2
70.5
0.6
Water-insolubles
30.1
28.8
29.4
-
``
Table 7. Product distributions of the the water-soluble and water-insoluble fractions and the whole bio-oil
Compound
Run 1
Run 2
Average
Standard deviation
WIF
Whole
WSF
WIF
Whole
WSF
WIF
Whole
WSF
WIF
Whole
GC detectables
29.45
17.88
25.82
27.06
17.57
27.32
28.25
17.72
26.82
1.19
0.15
0.25
Formaldehyde
4.45
-
3.48
4.19
0.27
5.06
4.32
-
4.27
0.13
-
0.79
Glycolaldehyde
7.72
1.00
5.19
7.90
0.79
7.05
7.81
0.90
6.12
0.09
0.11
0.93
Methyl glyoxal
3.08
-
2.76
2.50
0.20
2.19
2.79
-
2.47
0.29
-
0.28
Acetol
2.89
0.51
2.29
2.27
0.51
1.91
2.58
0.51
2.10
0.31
0.00
0.19
3-hydroxypropanal
1.09
0.34
0.80
1.10
0.33
0.77
1.09
0.33
0.79
0.01
0.00
0.02
0.29
0.17
0.16
0.35
0.12
0.11
0.32
0.14
0.14
0.03
0.03
0.02
1.05
0.40
0.87
1.03
0.37
0.91
1.04
0.38
0.89
0.01
0.01
0.02
Furfural
0.85
0.85
1.00
0.79
0.81
0.73
0.82
0.83
0.87
0.03
0.02
0.14
2-furanmethanol
0.23
0.21
0.23
0.22
0.21
0.23
0.22
0.21
0.23
0.00
0.00
0.00
2-methyl-2-
0.10
0.09
0.11
0.09
0.08
0.09
0.10
0.09
0.10
0.01
0.01
0.01
1.06
0.42
0.81
0.91
0.37
0.75
0.99
0.39
0.78
0.07
0.03
0.03
0.16
0.17
0.19
0.17
0.118
0.19
0.16
0.17
0.19
0.00
0.00
0.00
Dimethoxytetrahydrofur
an
2-hydroxy-3oxobutanal
cyclopenten-1-one
2-hydroxy-2cyclopenten-1-one
5-methyl furfural
62
WSF
``
Table 7. (continued)
Compound
3-methyl-2-
Run 1
WSF
WIF
Run 2
Whole
WSF
WIF
Average
Whole
WSF
WIF
Standard deviation
Whole
WSF
WIF
Whole
0.09
0.08
0.10
0.08
0.06
0.09
0.09
0.07
0.10
0.01
0.01
0.00
2(5H)-furanone
0.72
0.40
0.68
0.31
0.36
0.59
0.51
0.38
0.63
0.21
0.02
0.05
4-hydroxy-5,6-dihydro-
1.14
0.51
1.00
1.40
0.59
1.11
1.27
0.55
1.05
0.13
0.04
0.06
0.41
0.35
0.40
0.32
0.29
0.30
0.37
0.32
0.35
0.05
0.03
0.05
Phenol
0.09
0.11
0.09
0.10
0.11
0.08
0.09
0.11
0.09
0.01
0.00
0.01
2-methoxyphenol
0.13
0.12
0.19
0.15
0.28
0.07
0.14
0.20
0.13
0.01
0.08
0.06
2-methoxy-4-
0.04
0.25
0.10
0.06
0.26
0.09
0.05
0.25
0.09
0.01
0.00
0.00
0.18
0.13
0.14
0.15
0.13
0.16
0.17
0.13
0.15
0.02
0.00
0.01
0.25
0.59
0.23
0.14
0.48
0.19
0.20
0.54
0.21
0.05
0.06
0.02
Eugenol
0.00
0.23
0.07
0.00
0.19
0.07
0.00
0.21
0.07
0.00
0.02
0.00
5-hydroxymethyl
0.65
0.24
0.40
0.38
0.24
0.37
0.51
0.24
0.38
0.13
0.00
0.02
cyclopenten-1-one
(2H)-pyran-2-one
Methylcyclopentenolone
4-ethyl-2methoxyphenol
2-methoxy-4vinylphenol
furfural
63
methylphenol
``
Table 7. (continued)
Compound
Run 1
WSF
WIF
Run 2
Whole
WSF
WIF
Average
Whole
WSF
WIF
Standard deviation
Whole
WSF
WIF
Whole
Isoeugenol
0.09
1.11
0.41
0.07
1.00
0.36
0.08
1.05
0.38
0.01
0.06
0.02
4-methyl-2,6-
0.25
0.77
0.41
0.23
0.85
0.38
0.24
0.81
0.40
0.01
0.04
0.01
Vanillin
0.15
0.26
0.17
0.10
0.25
0.15
0.12
0.25
0.16
0.03
0.01
0.01
4-ethyl-2,6-
0.08
0.34
0.14
0.07
0.35
0.08
0.08
0.34
0.11
0.00
0.00
0.03
0.00
0.14
0.07
0.00
0.12
0.06
0.00
0.13
0.07
0.00
0.01
0.01
0.15
1.07
0.43
0.16
1.19
0.43
0.16
1.13
0.43
0.00
0.06
0.00
0.13
0.53
0.22
0.06
0.55
0.19
0.10
0.54
0.20
0.03
0.01
0.01
0.15
2.09
0.72
0.19
2.06
0.73
0.17
2.07
0.73
0.02
0.02
0.00
0.26
0.57
0.26
0.17
0.52
0.22
0.21
0.54
0.24
0.05
0.03
0.02
0.11
0.29
0.15
0.10
0.22
0.17
0.10
0.25
0.16
0.01
0.04
0.01
0.10
0.69
0.14
0.11
0.63
0.11
0.10
0.66
0.13
0.01
0.03
0.02
dimethoxyphenol
dimethoxyphenol
4-hydroxy-3methoxyacetophenone
vinylphenol
4-allyl-2,6dimethoxyphenol
2,6-dimethoxy-4-(1propenyl)phenol
4-hydroxy-3,5dimethoxybenzaldehyde
4-hydroxy-3,5dimethoxyacetophenone
Sinapialdehyde
64
2,6-dimethoxy-4-
``
Table 7. (continued)
Compound
Run 1
WSF
WIF
Run 2
Whole
WSF
WIF
Average
Whole
WSF
WIF
Standard deviation
Whole
WSF
WIF
Whole
Levoglucosan-furanose
0.40
1.06
0.42
0.35
1.21
0.36
0.38
1.14
0.39
0.02
0.08
0.03
1,4-benzenediol
0.15
0.08
0.13
0.12
0.18
0.10
0.13
0.13
0.11
0.02
0.05
0.02
1,4;3,6-dihydro-α-D-
0.31
0.09
0.21
0.31
0.06
0.25
0.31
0.08
0.23
0.00
0.01
0.02
-
0.34
0.10
-
0.39
0.11
-
0.36
0.10
-
0.03
0.00
0.06
0.03
0.05
0.05
0.02
0.04
0.05
0.03
0.04
0.01
0.00
0.01
IC detectables
11.06
1.91
7.09
10.19
1.69
6.94
10.62
1.80
7.01
0.43
0.11
0.07
Acetic acid
7.75
1.37
5.28
7.11
1.30
5.38
7.43
1.33
5.33
0.32
0.04
0.05
Propanoic acid
0.36
0.04
0.21
0.28
0.06
0.11
0.32
0.05
0.16
0.04
0.01
0.05
Glycolic acid
1.71
0.26
0.79
1.59
0.11
0.50
1.65
0.19
0.64
0.06
0.07
0.14
Formic acid
1.24
0.24
0.81
1.21
0.22
0.95
1.22
0.23
0.88
0.01
0.01
0.07
21.73
1.69
14.44
21.92
2.23
15.86
21.82
1.96
15.14
0.09
0.27
0.71
Cellobiose
0.10
0.00
0.10
0.39
0.00
0.20
0.25
0.00
0.15
0.14
0.00
0.05
Cellobiosan
1.52
0.04
1.19
1.65
0.04
1.20
1.58
0.04
1.20
0.07
0.00
0.00
Xylose
1.22
0.05
0.91
1.22
0.05
0.90
1.22
0.05
0.90
0.00
0.00
0.00
Maltosan
2.37
0.16
1.69
2.23
0.16
1.66
2.30
0.16
1.67
0.07
0.00
0.02
Levoglucosan
8.12
0.43
5.14
6.60
0.43
4.92
7.36
0.43
5.03
0.76
0.00
0.11
glucopyranose
Methanol
Acetaldehyde
65
HPLC detectables
``
Table 7. (continued)
Compound
Unidentified
Run 1
WSF
WIF
Run 2
Whole
WSF
WIF
Average
Whole
WSF
WIF
Standard deviation
Whole
WSF
WIF
Whole
8.40
0.30
5.41
9.83
1.55
6.98
9.11
0.92
6.19
0.71
0.62
0.78
28.00
23.73
26.71
29.07
21.74
27.31
28.53
22.73
27.01
0.54
1.00
0.30
28.00
23.73
26.71
29.07
21.74
27.31
28.53
22.73
27.37
0.54
1.00
0.66
Moisture water
-
-
13.49
-
-
13.84
-
-
13.66
-
-
0.18
Dehydration water
-
-
13.23
-
-
13.47
-
-
13.35
-
-
0.12
-
54.39
16.35
-
56.82
15.76
-
55.60
16.07
-
1.21
0.28
-
54.39
16.35
-
56.82
15.76
-
55.60
16.07
-
1.21
0.28
90.24
-
90.41
88.24
-
93.19
89.24
-
91.80
1.00
-
1.39
oligosaccharides
Karl-Fischer detectables
Water
Others
Total
All numbers in wt% based on each bio-oil. WSF stands for water soluble fraction of bio-oil. WIF stands for water insoluble fraction of
bio-oil. Whole stands for whole bio-oil. Theoretical Whole stands for calculated composition of the whole bio-oil, based on yields of
water-soluble and water-insoluble fractions and its distribution in the whole bio-oil. Unidentified oligosaccharides were quantified by
a subtraction of characterized sugars from total sugars estimated by acid-hydrolysis technique. Dehydration water was quantified by a
subtraction of moisture water in feedstock from water content of the whole bio-oil (by Karl Fischer).
66
Pyrolytic lignin
67
Figure 1. Schematic diagram of the fluidized bed reactor system.
1.0
Y = -0.931X + 1.923
2
R = 85.9
FID response factor
0.8
0.6
0.4
0.2
0.0
1.0
1.2
1.4
1.6
1.8
2.0
Total molecular weight / Total carbon molecular weight
Figure 2. Correlation between MWtotal/MWcarbon and FID response to approximate response
factors of commercially unavailable cyclic compounds.
68
Run 1
Run 2
30.0
26.727.0
25.0
Yield, %
20.0
15.0
17.9
16.415.8
15.4
15.9
14.4
10.0
7.1 6.9
5.0
4.6 4.1
2.6 2.2
2.4 2.3
Furans
Cyclics Phenols LMWs Pyrolytic Water
lignin
0.0
Sugars
Acids
Figure 3. Product distribution of whole bio-oil
Run 1
Run 2
30.0
25.0
24.4 24.6
Yield, %
20.0
14.1
15.0
13.0
11.1 10.2
10.0
5.0
2.9 2.2
2.8 2.8
2.7 2.3
Furans
Cyclics
Phenols
0.0
Sugars
Acids
Figure 4. Product distribution of water-soluble fraction
LMWs
69
Run 1
Run 2
60.0
56.8
54.4
50.0
Yield, %
40.0
30.0
20.0
10.610.1
10.0
2.8 3.5
1.9 1.7
2.0 1.9
1.5 1.4
Sugars
Acids
Furans
Cyclics
3.0 2.8
0.0
Phenols
Figure 5. Product distribution of water-insoluble fraction
LMWs Pyrolytic
lignin
70
CHAPTER 4
MANIPULATION OF CHEMICAL SPECIES IN BIO-OIL USING IN SITU CATALYTIC
FAST PYROLYSIS IN BOTH BENCH-SCALE FLUIDIZED BED PYROLYZER AND
MICROPYROLYZER
Yong S. Choi, Kyong-Hwan Lee, Jing Zhang, Robert C. Brown, and Brent H. Shanks
Abstract
In situ catalytic fast pyrolysis (CFP) was conducted with base or acid catalysts in bench-scale
pyrolyzer. Complete mass balance was performed for the first time, allowing to quantitatively
investigate the catalytic impacts on the final bio-oil composition. Acid catalysts exhibited
relatively higher activities for decomposition of sugar and pyrolytic lignin, dehydration,
decarbonylation, and coke formation, as relative to base catalysts. Carbon balance revealed that a
significant amount of carbon was transformed from the bio-oil to coke during CFP. CFP was also
performed in micropyrolyzer and the results were compared with those in the bench-scale
reactor, for the purpose of scale-up. Different from the bench-scale pyrolyzer, base catalyst
provided higher activities in the above reactions in micropyrolyzer, and the discrepancies suggest
a rapid deactivation of base catalyst. Due to the decrease in the bio-oil yield during CFP,
significantly less energy was recovered in CFP than in control pyrolysis.
Introduction
Thermochemical conversion of lignocellulosic biomass via pyrolysis continues to attract
attention as a potential pathway to renewable transportation fuels and chemicals. Biomass fast
pyrolysis is defined as rapid thermal degradation of lignocelluloses in the absence of oxygen to
primarily generate liquid bio-oil with a lesser amount of non-condensable gases and solid char
[1]. Bio-oil is a complex mixture of over 300 oxygenated organic compounds [2], which can be
71
broadly classified into anhydrosugars, phenolics, furan/pyran derivatives, and C2 to C4 low
molecular weight compounds. Due to an abundance of oxygenated compounds, bio-oil has an
oxygen content level, typically 35-40 wt%, which is similar to that of the biomass feedstock.
This oxygen content generally precludes bio-oil from being directly utilized for fuels or
chemicals and imparts a lower heating value for the bio-oil and higher reactivity during storage.
Thus, for bio-oil to be used in fuel or chemical applications, oxygen elimination is necessary.
Significant work has been performed on bio-oil deoxygenation after condensation. Additionally,
deoxygenation of pyrolysis vapors, either in situ or ex situ processing has been examined.
During in situ upgrading processes, the pyrolysis vapors are immediately contacted with catalytic
material in the pyrolysis reactor, whereas in ex situ processing the pyrolysis vapors undergo
catalytic reactions in a downstream reactor.
Significant attention has been paid to zeolites and zeolite-like materials for biomass
catalytic pyrolysis. Highly acidic zeolites have proven to be quite effective in reducing oxygen
levels in bio-oil; however, the deoxygenation is accompanied by a large decrease in the amount
of bio-oil produced and by increases in aromatic hydrocarbons and coke. Williams et al.
performed catalytic fast pyrolysis of rice husks with a ZSM-5 in a fixed bed reactor, and reported
that the oxygen removal down to 8.1 wt% relative to 37.7 wt% for the non-catalytic pyrolysis
was achieved by dehydration, decarboxylation and decarbonylation, although it was
accompanied with a large sacrifice of the organics from 37.0 wt% to 3.8 wt% [3]. They also
found that not only mono-hydrocarbons but also polycyclic aromatics hydrocarbon (PAH),
known to be carcinogenic [4, 5], were produced. Carlson et al. claimed that catalytic pyrolysis of
carbohydrates with ZSM-5 generated more than 30 wt% (based on carbon yield) of aromatic
hydrocarbons; however, a large amount of PAH (over 50% PAH selectivity) including
72
naphthalene and indene, were also generated [6, 7]. Severe coke formation on the catalyst was
also observed when using zeolite for upgrading pyrolysis vapors. Over 35 wt% of the feedstock
carbon ended up in the form of coke over on ZSM-5, resulting in bio-oil carbon loss as well as a
rapid catalyst deactivation. To reduce coke deposition while maintaining a high deoxygenation
level, researchers have tested mesoporous zeolite-like materials such as Al-MCM-41, MCM-41,
SBA-15, and MSU-3 for catalytic pyrolysis [8-10]. However, the coke yield was at best only
slightly lower than the base zeolite case.
Relatively fewer studies have been conducted on metal oxides for biomass catalytic
pyrolysis, although the materials have been used in many catalytic processes due to their unique
properties [11]. Lu et al. performed catalytic pyrolysis with 6 different metal oxides in a
Pyroprobe-GC/MS, and claimed that calcium oxide (CaO) significantly reduced anhydrosugars
and removed organic acids with increases in phenol and cyclopentanones, while magnesium
oxide (MgO) reduced levoglucosan slightly and increased linear aldehydes [12]. More recently,
Lin et al. performed catalytic pyrolysis with CaO in a fluidized bed reactor, and reported a
reduction in oxygen levels in bio-oil from 39 wt% to 31 wt%, resulting from a large decrease in
compounds having high oxygen content such as levoglucosan, formic acid and acetic acid [13].
Instead, the amount furanic compounds, which contain lower oxygen content, increased.
Although acid and base catalysts have been used for biomass catalytic pyrolysis by
researchers, there has not been a systematic evaluation of the impact of acidity or basicity on the
final bio-oil composition. More importantly, a complete mass balance on all of products
generated during pyrolysis has not been performed in previous studies, thereby limiting the
insight that can be gained. Hence, in the current study complete product speciation analysis of
bio-oil from catalytic experiments was performed by integrated use of GC, HPLC, IC, elemental
73
analyzer and Karl Fischer, and the composition analyses were compared with that for noncatalytic fast pyrolysis, to quantitatively demonstrated the catalytic effects on the bio-oil
chemical composition. Pyrolysis of a feedstock pretreated with H2SO4 was also conducted to
explore the impact of the treatment on the resulting bio-oil quality. This study also investigates
how much carbon is lost from the bio-oil during catalytic pyrolysis by completing the carbon
balance, which commonly been neglected. By considering the bio-oil yield as well as its higher
heating value (HHV), we are able to compare total energy recovered from the feedstock in the
bio-oil product for each pyrolysis experiment. In addition to the bench-scale fluidized bed
pyrolyzer, catalytic pyrolysis was conducted in micropyrolyzer, to determine whether catalytic
performance under micropyrolysis conditions can be used to project results in the bench-scale
fluidized bed reactor.
Experimental
Materials
Red oak was ground and sieved to 250-500 µm and was used as the feedstock for all of
the pyrolysis experiments. For the catalytic experiments, four different catalysts were used: MgO
(Sigma-Aldrich), MgO synthesized from MgCO3 (Sigma-Aldrich), γ-Al2O3 (Fischer scientific),
silica-alumina (Sigma-Aldrich, SiO2:Al2O3 ratio of 9.4:1), and commercial fluidized cracking
catalyst, FCC A (BP America, zeolite to matrix surface area ratio of 4.2).
Catalyst preparation
Catalysts used in the study could be categorized as acidic or basic with fresh (and
regenerated) γ-Al2O3, silica-alumina, and FCC A representing acid catalysts, and low surface
area MgO (Low MgO) and high surface area MgO (High MgO) representing base catalysts. The
catalyst selections were made to investigate how acidity and basicity in catalyst systematically
74
influence the composition of the resulting bio-oil. Since the surface area of commercially
available MgO (Sigma-Aldrich), 1.2 m2/g, is two orders of magnitude lower than that of the
other catalysts, High MgO with a surface area of 161.5 m2/g, was synthesized by calcination of
MgCO3 in static air using a ramp of 10 °C/min and hold at 550 °C for 2 hours [14].
All catalysts were received in powder form and thus were pelletized, crushed, and sieved
into a desired range of 0.4-0.6 mm, to ensure good fluidization for the catalytic pyrolysis
experiments. By utilizing the density of each catalyst type values of U/Umf and deq/D for the 10
L/min of carrier gas rate were set between 2 and 3 and below 70 % [15], respectively, by
adjusting the particle size, where U is the superficial velocity at the given flowrate at 500 °C, Umf
the minimum velocity for fluidization, deq the volume-equivalent diameter of bubble, and D the
bed diameter. For the micropyrolyzer experiments, both the feedstock and catalysts (calcined at
520 °C for 1 hour) were ground, sieved to 38-90 µm (to avoid transport limitations), and
physically mixed prior to pyrolysis.
Catalyst characterization
The acidity/basicity and surface area of the catalytic materials were measured by
NH3/CO2-TPD and BET analysis, respectively. For measuring both the density and strength of
the acid/base sites, NH3/CO2-TPD was performed using a Micromeritics AutoChem 2920. First,
approximately 0.1 g of sample was calcined at 700 °C for 1 hour to eliminate impurities,
followed by an injection of NH3/CO2 into the sample until saturation was achieved. After the
saturation, the sample was heated at a rate of 10 °C/min to 700 °C, and simultaneously the
desorbed NH3/CO2 was recorded by a thermal conductivity detector (TCD). The desorbed
amount of NH3/CO2 is proportional to levels of acid/base sites in the sample, respectively. The
specific surface areas were measured by nitrogen physisorption (Micromeritics ASAP 2020) and
75
calculated using the Brunauer-Emmett-Teller (BET) method. Approximately 0.1 g of sample was
weighed, and degassed at 350 °C for 5 hours to remove physisorbed species on the surface of the
sample. Nitrogen adsorption at relative pressure, P/P0, between 0.05 and 0.25 was used to
calculate the BET surface area.
Preparation of acid-infused red oak
Approximately 500 g of red oak in a size range of 250-500 µm was weighed and mixed
with sulfuric acid (0.4 wt% of red oak) and distilled water. The mixture was stirred for 20 min
followed by drying at 50 °C for 5 days. The resulting moisture content of 5.7 wt%, which was
30% lower than untreated red oak, was determined by proximate analysis. It is likely that the
decrease in moisture led to slight reductions observed for hydrogen and oxygen with the treated
feedstock.
Pyrolysis experiments
Catalyst activation
Prior to reaction, the 100 g of catalyst particles in the pyrolysis reactor were activated at
500 °C for at least 1 hour in 10 L/min of nitrogen. After activation, cyclones downstream of the
reactor were weighed to determine whether a significant amount of catalyst particles had
elutriated at the specific nitrogen flow rate. For the micropyrolyzer, catalysts were activated prior
to reaction in a muffle furnace in nitrogen environment at 520 °C for 1 hour.
Fluidized bed pyrolysis
A bench-scale bubbling fluidized bed reactor was used to conduct the catalytic fast
pyrolysis experiments. A detailed description of the pyrolyzer can be found elsewhere [16]. The
catalyst particles were used as the fluidizing medium for the catalytic experiments, with the
appropriate particle size determined so that the particles were fully fluidized without elutriating.
76
There were two types of carbon solids generated during the catalytic pyrolysis, which were
called char and coke. In the study, char was defined as the amount of solid carbonaceous
particles collected in the cyclones and in the reactor. Char was separated from the catalyst
particles by sieving. For measuring coke deposited on the catalyst, the post-reaction catalyst was
heated with 140 L/min of air, and the weight difference before and after this oxidation were
defined as the amount of coke. The regeneration of used Al2O3 was performed using this same
oxidation method.
Bio-oil analysis
Immediately after completion of a pyrolysis reaction run the three sets of bio-oil samples
(whole bio-oil, water-soluble fraction, and water-insoluble fraction) were prepared and were
individually analyzed for use in the overall mass balance. A detailed description of the
preparation methodology for each of the bio-oil samples was provided previously [16]. Complete
analysis of the bio-oil samples is difficult due to its complex mixture of >200 organic
compounds. Five different analytical techniques were employed in the current study to
quantitatively measure as many product species as possible. Full analysis of the bio-oil samples
was conducted by gas chromatography (GC), high-performance liquid chromatography (HPLC),
ion chromatography (IC), elemental analyzer, and Karl-Fischer. Details of each of the employed
analytical techniques was described previously [16].
Micropyrolysis
Micropyrolysis was conducted using a single shot micropyrolyzer (model 2020iS,
Frontier Laboratory, Japan) connected to a GC-MS/FID. For the catalytic experiments,
approximately 800 µg of a 1:1 wt mixture of the feedstock and catalyst was placed in a sample
cup and pyrolyzed at 500 °C. The resulting pyrolysis vapors were quantified by GC-MS/FID and
77
non-condensable gases (CO and CO2) by a De-Jaye gas analyzer equipped with an IR detector.
The solid residue in the sample cup post-pyrolysis was gravimetrically analyzed for char/coke
yield. The mass spectrometer (MS) and flame ionization detector (FID) was employed for
identification and quantification, respectively. The GC column used was Zebron ZB-1701 coated
with 14 % cyanopropylphenyl and 86 % dimethylpolysiloxane with dimensions of 60 m×0.25
mm ID×0.25 µm film thickness. The column was heated at a rate of 5 °C/min from 35 °C to 300
°C followed by a 4 min dwell at the final temperature. The volatiles generated in the reactor were
immediately swept by helium carrier gas (100 mL/min) and introduced to the column through an
injector maintained at 300 °C using a 100:1 split ratio. Peak assignment to a compound was
performed using the NIST mass spectra library and literature values [17-20] with quantification
from the respective FID calibration curve for each compound.
Results and discussions
Catalyst properties
The acid and base properties, specific surface area and pore properties of catalysts tested
are summarized in Table 1. The silica-alumina had the largest number of acid sites as well as the
strongest acidity. For the γ-Al2O3, both fresh and regenerated samples were tested and the
number of acid sites only decreased a small amount upon regeneration. Interestingly, the
regenerated Al2O3 gave more strongly bound NH3 suggesting a slightly stronger acidity. The
FCC A material had a small number of acid sites with rather weak acidity and the lowest surface
area among the acid catalysts. As can be seen from the CO2-TPD and surface area results, the
High MgO had an order of magnitude greater number of base sites than the Low MgO most
likely attributable to the increased surface area for the High MgO.
78
Catalytic pyrolysis of biomass in bench-scale reactor
Catalytic effects on overall mass balance
Duplicate runs of red oak fast pyrolysis with inert silica sand as the fluidizing medium
were performed in our previous study [16] and these results were used as the baseline in the
current study. Detailed overall mass balances including the non-catalytic control and catalytic
experiments are given in Fig. 1. The total mass closures for all of the pyrolysis experiments were
between 85 % and 94 %. The unaccounted for mass was likely attributed to a small amount of
bio-oil left in pipes and fittings, which were not measured in the study, and from errors
associated with measuring non-condensable gases. As can be seen from Fig. 1, a decrease in biooil yield was accompanied with increases in coke and non-condensable gases with relatively
constant char yield when using the catalysts. In the bio-oils, a significant amount of the organics
decreased and water increased with the catalysts, possibly due to catalyzed C-C cleavage,
decarbonylation, and dehydration reactions. It was observed with catalytic pyrolysis that a
decrease in the yield of bio-oil was accompanied by increases in coke (up to 10.6 wt%) and noncondensable gases (up to 18.9 wt%), which was the most pronounced with the acidic catalysts.
The basic Low MgO produced a small yield of coke (1.9 wt%), probably due to its low surface
area, while the High MgO formed as high as 8.2 wt%. A slight increase in dehydration water
was observed with the base catalysts, which was consistent with basic catalysts promoting
dehydration reactions [21]. The product distribution was altered to only a small extent with the
FCC A catalyst likely due to the small number of acidic sites with relatively weak acidity.
Interestingly, the amount of char generated was relatively consistent throughout all the runs
(except for the acid-infused case), suggesting that char formation was not catalyzed by acidic or
basic catalysts. The highest char generation (29.6 wt%) was observed for pyrolysis of the acid-
79
infused biomass; however, it appeared that reactions besides typical char-forming reactions
appeared to occur. This speculation was based on two observations: 1) char was collected mainly
in the reactor rather than cyclones where most of the char was collected in the other pyrolysis
experiments, and 2) the collected char in a reactor was agglomerated with the sand particles,
suggesting heat and mass transfer issues.
Catalytic effects on water-soluble and water-insoluble fractions of bio-oil
Table 2 presents the mass percentages of the water-soluble and water-insoluble fractions
in the whole bio-oil for all of the pyrolysis runs. In general, water-insolubles primarily contain
lignin-derived compounds such as pyrolytic lignin (oligomers of phenolic compounds), whereas
the water-solubles mainly consisted of carbohydrate-derived species. For the non-catalytic
control case, the water-insolubles made up approximately 29 wt% of the whole bio-oil. The
water-insolubles portion decreased upon introduction of the catalysts except for FCC A. This
decrease in water-insolubles was more evident for the acidic catalysts rather than the basic
catalysts, suggesting that acidic catalysts were more active in cracking reactions of the pyrolytic
lignin precursors. In the case of the acid-infused sample only 11 wt% of the whole bio-oil was
water-insolubles, which is discussed in more detail in a following section (3.2.4.1). The FCC A
catalyst, which had the lowest acid levels and surface area, imparted little manipulation of the
water-solubles and water-insolubles ratio.
Catalytic effects on non-condensable gases
The non-condensable gases increased for of the solid catalyst runs, whereas a reduction in
gases was observed for the acid-infused material (see Fig. 1 for overall amount of the gases and
Table 3 for the speciation). The most significant shift in speciation was an increase in CO
content on the products gases, which would be expected to mainly be contributed by
80
decarbonylation. The CO content in the non-condensable gases was reduced for the acid-infused
material suggesting suppression of the decarbonylation reaction (over 50 %). Based on the
distribution of the gases, the basic catalysts had higher activity for dehydrogenation and
exhibited a slightly lower level of decarbonylation activity, compared with acidic catalysts.
Unfortunately, none of the catalysts tested enhanced the relative level of decarboxylation
activity, which optimize oxygen removal by eliminating two oxygen atoms with one carbon
atom. Typically the desire is to achieve some amount of deoxygenation during catalytic
pyrolysis, so the fact that decarbonylation appeared to be enhanced but not decarboxylation was
undesirable as it would have a negative impact on the weight of products recovered in the bio-oil
organics.
Catalytic effects on product distributions
Catalytic effects on overall product distributions based on wet feedstock
As shown in Fig. 2, the decomposition of sugars and pyrolytic lignin with concomitant
formation of dehydration products was pronounced with the acidic catalysts. As consistent char
yields and phenolic and LMW compounds decreases were observed for all of the solid acid runs,
it appeared that the lost sugars and pyrolytic lignins were balanced with increased coke
formation on the catalysts. This transformation of the organics from bio-oil to coke is not
positive as it would lead to catalyst deactivation and lower recovery of carbon species in the biooil. The extent of dehydration was proportional to the acid site densities for the catalysts. For
example, a large amount of water was generated with the silica-alumina, 27.8 wt%, which had
the most acid sites, while the degree of dehydration with FCC A, which had the least number of
acid sites, was minimal. The basic catalysts also were found to increase the extent of
dehydration, but the extent of the reaction was lower than with the acidic catalysts. The
81
regenerated Al2O3 had similar activity for sugar decomposition and dehydration as the fresh
Al2O3. However, diminished pyrolytic lignin conversion was observed with the regenerated
catalyst, which could be due to loss of micropores through the regeneration process. In fact, the
micropores were not recovered upon regeneration as none were measured post-regeneration
compared to a micropore volume of 0.00349 cm3/g for the fresh material (Table 1).
In contrast to the solid acid catalysts, which catalyzed sugar decomposition, pyrolysis of
the acid-infused feedstock without a solid catalyst generated 20 % more sugars, particularly
levoglucosan, and the increase was accompanied by a decrease in the amount of the LMW
compounds. Kuzhiyil et al. previously reported switchgrass passivated with H2SO4 generated
approximately 16 times more anhydrosugar during pyrolysis, with a significant decrease in
LMW compounds [22]. It was speculated that mineral-catalyzed ring-opening reactions were
inhibited by forming thermally stable salts with indigenous biomass minerals thereby allowing
for more glycosidic bond cleavage. In the current work relative to the control case, the char yield
greatly increased from 11.6 wt% to 29.6 wt% with reductions in the pyrolytic lignin (from 9.8
wt% to 3.5 wt%) and phenols (from 2.9 wt% to 0.6 wt%). Based on the observation of a marked
decrease in lignin-derived products along with an increase in the sugars, the char increase
appeared to come primarily from a decrease in the lignin-derived products. This result was
consistent with the markedly lower water-insolubles found for the passivated run, which
predominantly consisted of pyrolytic lignin and phenolic compounds (Table 2).
For the basic catalysts, activities were observed for sugar decomposition, fragmentation
to gases and dehydration; however, the overall extent of conversion was lower than with the
acidic catalysts. Although number of basic sites in the High MgO increased by an order of
magnitude relative to the Low MgO due its higher surface area, the High MgO had only slight
82
increases in the apparent reaction product species compared to the Low MgO. As the High MgO
had a large in coke formation, it appeared that the higher surface area was mitigated by rapid
deactivation during pyrolysis. Based on the yields of pyrolytic lignin for basic and acidic
catalysts, the expected result of pyrolytic lignin conversion being promoted by acidity rather than
basicity was observed.
Catalytic effect on product distributions for the water-soluble fraction of bio-oil
As shown in Fig. 3, the water-soluble fractions were primarily composed of sugars,
organic acids, and LMW compounds, with only trace amounts of compounds with low water
solubility such as phenolic compounds. The product speciation of the water-soluble fractions of
the product bio-oil followed similar trends to those of whole bio-oil. Less sugar and more water
were found with the acidic catalysts than with the basic catalysts. A relatively low degree of
product modification was observed when using FCC A, which had the least number of acidic
sites.
Catalytic effect on product distributions for the water-insoluble fraction of bio-oil
The species breakdowns for bio-oil water-insoluble fraction are shown in Fig. 4. A trace
amount of water-solubles compounds such as sugars, acids, furans, and LMW compounds
derived from carbohydrates were recovered in the water-insoluble fraction. This was due to
difficulty in achieving complete phase separation for the whole bio-oil. With the acidic catalysts
(except for FCC A) more phenolic compounds remained in the water-insolubles compared to the
control case. The fresh Al2O3 gave the highest recovery of phenolic compounds (16.6 wt%).
Similar to the trend for the whole bio-oil, less sugar was observed in the water-insolubles with
the acidic catalysts and slightly more sugars with the acid-infused run. Interestingly, the product
distribution for the acid-infused material was dramatically different than that of the solid catalyst
83
runs. For example, a marked decrease in the phenolic compounds with increases in sugar and
furans was observed in the passivated run products.
Carbon balance
By calculating the overall carbon balance it was possible to determine how much carbon
was sacrificed relative to the degree of bio-oil deoxygenation. During catalytic pyrolysis three
different reactions would remove oxygen, decarboxylation, decarbonylation, and dehydration,
with either carbon or hydrogen consumption required. As seen in Table 4, more carbon ended up
in the form of coke and non-condensable gases with catalytic fast pyrolysis than with the control
case. For the basic catalysts, the carbon lost from the bio-oil to coke was more severe for the
High MgO (17.9 wt%) than the Low MgO (4.2 wt%), probably due to the higher surface area of
the High MgO. Similarly, the acid catalyst runs (except for FCC A) led to a significant amount
of carbon loss in the form of coke, so the amount of carbon that was recovered in the bio-oil was
decreased significantly. For the acid-infused material, the maximum carbon loss to char was
observed relative to any of the solid catalysts with only 32.9 wt% of the feedstock carbon
remaining in the bio-oil. In examining the carbon balances, carbon from species that would have
been observed in water-insoluble fraction contributed more to the char formation, relative to
species that would have been in the water-soluble fraction. This observation was further evidence
of pyrolytic lignin conversion to char for the acid-infused feedstock.
Higher heating value (HHV) of feedstock and bio-oil product
Since high levels of oxygen in the bio-oil reduce the HHV for the bio-oil, it is important
to monitor oxygen content in the product. As can be seen Table 5, the control bio-oil contained
35.3 wt% oxygen, which was approximately a 20 % decrease from the dry feedstock. The fresh
and regenerated Al2O3 showed reduction in oxygen content to 30.7 wt% and 32.8 wt%,
84
respectively, thereby leading to only small increases in the HHV for the bio-oil product from
21.9 MJ/kg to 23.2 MJ/kg and 22.5 MJ/kg, respectively. In contrast, no diminishment in oxygen
content was achieved when using the basic catalysts, which was consistent with lower sugar
decomposition and dehydration activities. In addition to oxygen content, hydrogen content in the
bio-oil product influences the HHV of bio-oil, so hydrogen was monitored as well. While the
fresh Al2O3 catalyst reduced oxygen content by approximately 13 %, the HHV only increased by
5 % due to a hydrogen loss from 5.2 wt% to 4.3 wt%. For the acid-infused material, the
significantly lower carbon recovery (due to char formation) made the resulting bio-oil HHV
lower than with the control case.
Total energy in bio-oil relative to the feedstock
Although there was a slight increase in the bio-oil HHVs for several of the catalysts, the
energy retention in the bio-oil product versus the starting feedstock needs to be considered. This
comparison needed to be made based a common amount of feedstock since the catalytic trials
resulted in significant decreases in the total amount of bio-oil recovered due to the significant
consumption of carbon and hydrogen during the catalytic reactions. In Table 6, the amount of
energy in the bio-oil product generated from the pyrolysis of 1 kg of feedstock was calculated for
all of the experiments by multiplying bio-oil HHV with the yield of the bio-oil organics. Due to
the significant decrease in organics in the bio-oil, all of catalytic pyrolysis experiments gave
lower recovered energy than with the control case. Among the catalytic runs, catalysts having
lower acidity or basicity such as FCC A and Low MgO maintained the higher energy recoveries
than catalysts having higher acidity or basicity such as Al2O3, silica alumina, and High MgO.
This result clearly demonstrates the importance of considering the energy content of the bio-oil
as well as the overall bio-oil yield when comparing catalytic pyrolysis to thermal pyrolysis.
85
Catalytic pyrolysis in the micropyrolyzer
A significant amount of literature reports a use of micropyrolyzers to evaluate the use of
catalytic materials during pyrolysis. Therefore, experiments were performed in a micropyrolyzer
to compare their results with those obtained and discussed above for the bench-scale fluidized
bed pyrolyzer. By performing complete mass balances with product speciation it was possible to
determine the degree of the similarities and differences in product distributions between the
pyrolyzers, which is important for translating micropyrolyzer results to continuous pyrolyzers.
Catalytic effects on overall mass balance
For the micropyrolyzer work, triplicate runs were conducted for all of the catalytic and
non-catalytic experiments. As shown in Table 3, mass balances (based on a wet feedstock) from
the experiments resulted in mass closures of approximately 80-90 %, which was slightly lower
than those for the bench-scale reactor. The lower mass balance closure was likely attributed to
the determination of dehydration water as well as the oligomers derived from carbohydrates
and/or lignin and non-condensable gases besides CO and CO2, which could not be detected in the
given experimental setup. A similar overall mass balance trend was observed for the
micropyrolyzer as with the bench-scale pyrolyzer. In the micropyrolyzer, the GC-detectables are
the species that upon condensation would be the bio-oil organic product. With the solid catalysts,
a decrease in yield of GC-detectables was accompanied by increases in amount of gases and
char/coke. However, the relative yield changes for the catalysts were not consistent between the
two pyrolyzers. The largest changes in the overall yields were found for the basic catalyst, High
MgO, rather than the acidic catalysts. For the experiments, the bench-top pyrolyzer was being
used with a continuous feed and the resulting products represented a quasi-steady state
performance whereas the micropyrolyzer was a batch process. As such, the micropyrolyzer
86
essentially provided information on the initial performance of a catalyst. Therefore, the yield
discrepancies for the different catalysts would suggest that basic catalysts were more prone to
lose activity with time than were the acidic catalysts. With the acid-infused material, the
micropyrolysis experiments gave a significantly lower amount of char generation than seen with
the continuous pyrolyzer. This difference was likely due to the significant char agglomeration
with the sand particles in the reactor that was observed with the continuous process, which would
not occur in the micropyrolyzer. However, for both reactor systems the increase in char was
correlated with the loss of lignin-derived species in the products. For the micropyrolyzer, the
amount of char increased from 11.5 wt% to 16.7 wt% without a sacrifice of the GC-detectables,
which likely implied that the pyrolytic lignin (not detected in the experimental setup) mainly
contributed to the char increase.
Catalytic effect on overall product distributions based on wet feedstock
As shown in Fig. 5, all of the solid catalysts promoted sugar decomposition with the basic
catalyst showing higher activity for fragmentation to non-condensable gases than with the acidic
catalysts. This trend was counter to that observed with the bench-scale pyrolyzer where the basic
catalysts gave lower activities for the sugar decomposition reactions. Again, the discrepancy in
the sugar, LMW compounds and gases trends with the basic catalyst between the two reactors
could be due to catalyst deactivation in the continuous pyrolyzer. As with the continuous
pyrolyzer, the FCC A with its low acid site density and relatively weak acidity in the
micropyrolyzer did not substantially alter the product distribution, which suggested that its low
overall activity in continuous process primarily resulted from catalyst properties rather than from
its deactivation.
87
As expected from the bench-scale unit, sugar formation was enhanced during
micropyrolysis of the acid-infused feedstock although the increase (by a factor of 3.6) was
significantly larger in the batch unit. This phenomenon could be due to modified transport
properties in the larger reactor system. The significant agglomeration observed in the fluidized
bed reactor, which was not replicated in the micropyrolyzer batch experiments, likely altered
both the heat and mass transfer characteristics within the reactor. Comparison of these results
suggests that problems associated with char agglomeration must be resolved to maximize sugar
production in a continuous pyrolysis reactor system.
Carbon balance
In the micropyrolyzer there was no way to differentiate char and coke, so they had to be
considered in aggregate. The carbon distributions in the form of organics, char/coke, and noncondensable gases for the micropyrolyzer runs are shown in Fig. 6. Due to undetectable
oligomers such as pyrolytic lignin, the amount of carbon present in the organics (effectively the
bio-oil minus water) in Fig. 6 was obtained by difference. Similar to bench-scale pyrolyzer, a
severe amount of carbon was lost in micropyrolyzer in the form of char/coke and noncondensable gases with the solid catalysts. This observation was most prominent with the High
MgO (42 wt% carbon in the organics) and silica alumina (46 wt% carbon in the organics)
relative to other catalysts. Consistent with all of the other results, the FCC A did not significantly
alter the carbon distributions. In the case of acid-infused feedstock run, far more carbon was
recovered in the organics (63.5 wt%) in the micropyrolyzer than in the bench-scale pyrolysis
system (32.9 wt% carbon recovery).
88
Conclusions
In the bench-scale pyrolyzer, acid catalysts showed high activities in sugar
decompositions, coke formation and gas fragmentation, while base catalyst was more active in
those reactions in micropyrolyzer. The discrepancies imply that base catalyst is more likely
deactivated with time, than acid catalyst. For the micropyrolysis of the acid-infused feedstock,
more anhydrosugars were generated as relative to the bench-scale pyrolyzer. The difference
indicates that heat/mass transfer limitations in the continuous pyrolyzer must be resolved for the
continuous production of anhydrosugars. Due to the decrease in the bio-oil yield during CFP,
energy recoveries were lower in CFP than in control pyrolysis.
References
1.
Huber, G.W. and J.A. Dumesic, An overview of aqueous-phase catalytic processes for
production of hydrogen and alkanes in a biorefinery. Catalysis Today, 2006. 111(1–2): p.
119-132.
2.
Mohan, D., C.U. Pittman, and P.H. Steele, Pyrolysis of Wood/Biomass for Bio-oil: A
Critical Review. Energy & Fuels, 2006. 20(3): p. 848-889.
3.
Williams, P.T. and N. Nugranad, Comparison of products from the pyrolysis and
catalytic pyrolysis of rice husks. Energy, 2000. 25(6): p. 493-513.
4.
Barfknecht, T., et al., Soot and mutation in bacteria and human cells. Chemical Analysis
and Biological Fate: Polynuclear Aromatic Hydrocarbons (M. Cooke, and AJ Dennis,
Eds.), Battelle Press, Columbus, OH, 1981: p. 231-242.
5.
Longwell, J.P., Polycyclic aromatic hydrocarbons and soot from practical combustion
systems, in Soot in Combustion Systems and its Toxic Properties1983, Springer. p. 37-56.
6.
Carlson, T.R., et al., Aromatic production from catalytic fast pyrolysis of biomass-derived
feedstocks. Topics in Catalysis, 2009. 52(3): p. 241-252.
7.
Carlson, T.R., et al., Production of green aromatics and olefins by catalytic fast pyrolysis
of wood sawdust. Energy & Environmental Science, 2011. 4(1): p. 145-161.
8.
Pattiya, A., J.O. Titiloye, and A.V. Bridgwater, Fast pyrolysis of cassava rhizome in the
presence of catalysts. Journal of Analytical and Applied Pyrolysis, 2008. 81(1): p. 72-79.
89
9.
Adam, J., et al., In situ catalytic upgrading of biomass derived fast pyrolysis vapours in a
fixed bed reactor using mesoporous materials. Microporous and Mesoporous Materials,
2006. 96(1): p. 93-101.
10.
Jackson, M.A., D.L. Compton, and A.A. Boateng, Screening heterogeneous catalysts for
the pyrolysis of lignin. Journal of Analytical and Applied Pyrolysis, 2009. 85(1): p. 226230.
11.
Bartholomew, C.H. and R.J. Farrauto, Fundamentals of industrial catalytic
processes2011: Wiley-AIChE.
12.
Lu, Q., et al., Catalytic upgrading of biomass fast pyrolysis vapors with nano metal
oxides: an analytical Py-GC/MS study. Energies, 2010. 3(11): p. 1805-1820.
13.
Lin, Y., et al., Deoxygenation of Bio-oil during Pyrolysis of Biomass in the Presence of
CaO in a Fluidized-Bed Reactor. Energy & Fuels, 2010. 24(10): p. 5686-5695.
14.
Bartley, J.K., et al., Simple method to synthesize high surface area magnesium oxide and
its use as a heterogeneous base catalyst. Applied Catalysis B: Environmental, 2012.
128(0): p. 31-38.
15.
Kunii, D. and O. Levenspiel, Fluidization engineering. Vol. 2. 1991: ButterworthHeinemann Boston.
16.
Choi, Y.S., et al., Detailed Characterization of Red Oak-Derived Pyrolysis Oil;
Integrated use of GC, HPLC, IC, GPC and Karl-Fischer. Journal of Analytical and
Applied Pyrolysis, 2014.
17.
Patwardhan, P.R., R.C. Brown, and B.H. Shanks, Understanding the fast pyrolysis of
lignin. ChemSusChem, 2011. 4(11): p. 1629-1636.
18.
Patwardhan, P.R., et al., Product distribution from fast pyrolysis of glucose-based
carbohydrates. Journal of Analytical and Applied Pyrolysis, 2009. 86(2): p. 323-330.
19.
Faix, O., et al., Thermal degradation products of wood: gas chromatographic separation
and mass spectrometric characterization of polysaccharide derived products. Holz als
Roh-und Werkstoff, 1991. 49(5): p. 213-219.
20.
Faix, O., D. Meier, and I. Fortmann, Thermal degradation products of wood. Gas
chromatographic separation and mass spectrometric characterization of monomeric
lignin-derived products. Holz als Roh-und Werkstoff, 1990. 48(7-8): p. 281-285.
21.
Lundeen, A.J. and R. VanHoozer, Selective catalytic dehydration. Thoria-catalyzed
dehydration of alcohols. The Journal of Organic Chemistry, 1967. 32(11): p. 3386-3389.
90
22.
Kuzhiyil, N., et al., Pyrolytic Sugars from Cellulosic Biomass. ChemSusChem, 2012.
5(11): p. 2228-2236.
23.
Demirbaş, A., Calculation of higher heating values of biomass fuels. Fuel, 1997. 76(5): p.
431-434.
Tables and figures
Table 1. Catalyst characterization
NH3 TPD,
mmol/g
Low MgO
High MgO
Fresh Al2O3
0.47
Regenerated Al2O3
0.45
Silica alumina
0.63
FCC A
0.11
Temp. at
maximum, °C
243.2
275.0
309.3
195.4
CO2 TPD,
mmol/g
0.73
17.40
-
BET surface
area, m2/g
1.2
161.5
318.2
234.5
519.1
176.4
t-plot micropore
volume, cm3/g
~0
0.00925
0.00349
~0
~0
0.06408
Table 2. Weight percentage for the water-soluble and water-insoluble fractions in the whole bio-oil
Water-insolubles
29.4
28.5
28.5
22.3
25.1
19.8
30.5
11.1
91
Control
Low MgO
High MgO
Fresh Al2O3
Regenerated Al2O3
Silica alumina
FCC A
Acid-infused
Water-solubles
70.5 ± 0.6
71.5
71.5
77.7
74.9
80.2
69.5
88.9
Table 3. Breakdowns of non-condensable gases
Control
H2
0.01±0.00
CH4
0.48±0.02
Low MgO
0.07
0.80
High MgO
0.04
0.63
Fresh Al2O3
0.03
0.83
Regenerated Al2O3
0.03
0.93
Silica alumina
0.01
0.77
FCC A
0.02
0.97
Acid-infused
0.42
All numbers in wt % based on feedstock.
C2H6
0.02±0.02
C2H4
0.16±0.01
CO
5.34±0.05
CO2
7.29±0.25
0.18
0.18
0.17
0.18
0.19
0.21
0.34
0.17
0.26
0.32
0.34
0.37
0.20
7.27
6.03
8.82
9.15
9.25
10.38
2.35
7.57
8.46
8.63
7.47
7.40
6.93
6.10
Total
13.30±0.1
7
16.23
15.32
18.75
18.07
17.95
18.86
9.28
CO/CO2
0.73
0.96
0.71
1.02
1.22
1.25
1.49
0.38
92
Table 4. Carbon balance
Control
Low MgO
High MgO
Fresh
Al2O3
35.0
20.8
14.2
Regenerated
Al2O3
39.1
22.7
16.4
Silica
alumina
30.1
16.2
13.9
FCC A
Acidinfused
32.9
26.3
6.6
Bio-oil
57.7
50.6
39.1
46.5
Water-solubles
36.5
29.8
20.4
23.5
Water21.2
20.8
18.7
23.0
insolubles
Char
19.4
16.8
14.7
17.9
16.0
15.9
19.8
45.9
Coke
4.2
17.9
17.8
19.7
22.0
11.4
Gas
10.5
13.5
12.4
15.5
15.4
15.3
16.4
7.0
Sum
87.6
85.1
84.1
86.2
90.2
83.3
94.1
85.8
All numbers in wt % based on total carbon in feedstock and obtained from elemental analysis. Coke is assumed to be pure carbon.
Water-solubles and water-insolubles include organics and water originated from feedstock and dehydration water.
Table 5. Composition and HHV of feedstock and whole bio-oil
Carbon, %
Hydrogen, %
Oxygen, %
HHV, MJ/kg
Red oak (dry)
50.0
6.3
43.6
19.0
Control
59.5±1.9
5.2±0.5
35.3±1.1
21.9±0.1
Low MgO
58.5
5.1
36.1
21.3
High MgO
51.5
7.7
38.5
22.3
Fresh Al2O3
64.9
4.3
30.7
23.2
Regenerated Al2O3
62.1
4.7
32.8
22.5
Silica alumina
57.1
4.8
38.0
20.1
FCC A
52.0
6.3
40.9
20.2
Acid-infused
50.7
7.3
38.9
21.5
All numbers in wt % based on dry whole bio-oil and obtained from elemental analysis. HHV was calculated by the formula, HHV =
{33.5[C] + 142.3[H] - 15.4[O] - 14.5[N]} x 10-2.
93
Table 6. Total energy generated in the form of bio-oil on the basis of 1 kg of feedstock
Total energy generated, MJ
Control
9.8 ± 0.2
Low MgO
8.3
High MgO
7.6
Fresh Al2O3
5.8
Regenerated Al2O3
6.5
Silica-alumina
4.1
FCC A
8.2
Acid-infused
6.6
Table 7. Overall mass balance in the micropyrolyzer
Feedstock
Total
moisture
Control
40.7±0.6
11.5±0.6**
21.9±2.4
8.4±0.1
82.5±0.7
High MgO
22.5±3.6
22.0±1.4
34.6±3.4
8.4±0.1
87.4±3.6
Fresh Al2O3
33.8±0.3
15.9±0.9
29.0±3.3
8.4±0.1
87.1±0.3
Silica alumina
32.4±0.4
21.9±1.2
25.6±2.3
8.4±0.1
88.2±0.4
FCC A
35.9±1.1
13.1±0.1
21.8±1.3
8.4±0.1
79.2±1.1
Acid-infused
40.6±1.4
16.7±0.3**
16.2±1.0
5.7±0.1
79.2±1.4
All numbers in wt % based on wet feedstock. In the cases of control and acid-infused experiments, only char amount was presented.
Feedstock moisture was measured by TGA.
GC-detectables
Char/coke*
Gas
94
Organics, bio-oil
13.2
0.0
11.6
16.4
15.4
1.9
10.5
15.5
8.2
9.2
19.9
Water, bio-oil
18.7
18.4
8.1
9.0
11.3
10.0
18.9
24.0
44.7
Low MgO
34.0
Coke
Gases
18.9
18.0
10.6
9.6
5.2
11.6
17.0
27.8
40.4
24.8
High MgO Fresh A₂lO₃ Reg. Al₂O₃
20.6
Silica
alumina
29.6
17.0
22.7
29.0
9.3
0.0
FCC A
30.5
Acid-infused
Figure 1. Overall mass balance in the bench-scale pyrolyzer. All numbers in wt % based on feedstock.
95
Control
39.1
Char
Sugars
Phenols
Non-condensable Gas
Acids
LMWs
Char
Furans
Water
Coke
Cyclics
Pyrolytic Lignin
35
30
Yield, %
25
20
15
10
5
96
0
Control
Low MgO
High MgO
Fresh Al₂O₃
Reg. Al₂O₃ Silica alumina
Figure 2. Product distribution of catalytic fast pyrolysis. All numbers in wt % based on feedstock.
FCC A
Acid-infused
97
80
Sugars
Acids
Furans
Cyclics
Phenols
LMWs
Water
70
60
Yield, %
50
40
30
20
10
0
Control
High MgO
Fresh Al₂O₃ Silica alumina
FCC A
Acid-infused
Figure 3. Product distribution of the water-soluble fraction. All numbers in wt % based on biooil.
98
Sugars
Acids
Furans
Cyclics
Phenols
LMWs
18
16
14
Yield, %
12
10
8
6
4
2
0
Control
High MgO
Fresh Al₂O₃ Silica alumina
FCC A
Acid-infused
Figure 4. Product distribution of the water-insoluble fraction. All numbers in wt % based on biooil.
99
Sugars
Phenolics
40
Acids
LMWs
Furans
Gas
Cyclics
Char/coke
35
30
Yield, %
25
20
15
10
5
0
Control
High MgO
Fresh Al₂O₃ Silica alumina
FCC A
Acid-infused
Figure 5. Product distribution of catalytic fast pyrolysis in the micropyrolyzer. All numbers in wt
% based on feedstock.
100
100
90
19.3
35.4
Carbon yield, %
80
70
25.1
14.7
20.3
60
25.9
14.7
10.6
62.9
63.5
FCC A
Acid-infused
18.1
22.7
50
22.4
36.3
40
30
66.0
54.6
42.0
20
45.5
10
0
Control
High MgO Fresh Al₂O₃
Organics
Gases
Silica
alumina
Char/coke
Figure 6. Carbon balance for micropyrolyzer. Carbon in organics was determined by difference.
101
CHAPTER 5
ACETIC ACID REMOVAL FROM PYROLYSIS VAPORS USING CaCO3
Yong S. Choi, Jing Zhang, and Brent H. Shanks
Abstract
Removal of organic acid in the pyrolysis vapor phase was investigated for the first time
in the constraint of almost complete pyrolysis product distribution, from which the selectivity
towards the acid removal could be evaluated. The acid removal was performed in a Tandem
micropyrolyzer coupled with a GC system for online product analysis. Broad range of
catalysts/adsorbents was tested, among which CaCO3 showed the best potential for acid removal.
Successful catalyst regeneration was achieved by in situ heat treatment on the postreaction
catalyst bed during which the adsorbed acid was simultaneously converted into ketone.
Extensive catalyst characterization and theoretical calculation suggested the acid removal
occurred only on the surface of CaCO3, probably through a monolayer adsorption. Mitigation on
mass loss of other products was achieved by using surface modified CaCO3 materials, resulting
in decent selectivity for acid removal. XPS analysis suggested the surface modification led to a
formation of a metal-carboxylate intermediate consisting both acetate and carbonate groups.
Incorporating the acetate group on CaCO3 led to its surface passivation thereby depressing side
reactions, as proven by inertness of calcium acetate to the pyrolysis vapors.
Introduction
Recently, sustainable routes to chemicals and fuels have been driven by researchers, due
to the depletion of fossil resources, increasing energy demand, and concerns over global
warming. Although alternative energy sources such as solar and wind are environmentally
102
friendly, high capital costs and low utilization rates make them less attractive as energy sources.
Additionally, the energy sources cannot replace transportation fuels and commodity chemicals
that have been produced by fossil resources. In contrast, biomass is a renewable energy source,
and has a potential to generate hydrocarbon fuels and chemicals. As a resource of renewable
carbon, lignocellulosic biomass consists of cellulose, hemicellulose, and lignin. There are two
main technologies of utilizing renewable carbon, which can be broadly categorized as
biochemical and thermochemical conversion routes[1]. Although ethanol generated
biochemically has been commercially available, there are huge challenges in that only sugar
portion of biomass was employed, leaving lignin as a byproduct of low value. In contrast, fast
pyrolysis, one of thermochemical technologies, converts a whole biomass including lignin into
liquid bio-oil that can be chemically upgraded into fuels and chemicals. Fast pyrolysis is defined
as a thermal decomposition at atmospheric pressure in oxygen-free environment. As bio-oil is a
complex mixture of over 300 organic compounds, the oil is considered renewable sources for the
production of fuels and chemicals[2, 3]. The bio-oil species can be broadly classified into
anhydrosugars, phenolics, furan/pyran derivatives, acids, and low molecular weight
compounds[4].
Among the bio-oil species, a good amount of carboxylic acids (7 wt%) is present, and
especially acetic acid is the main organic acid in bio-oil, thus making the oil is acidic (pH 23)[4]. The high acidity causes bio-oil to be less promising candidate for renewable resources. For
example, the acidity makes it difficult to store the oil due to its corrosivity, and more importantly
highly reactive acids may also participate in or promote reactions occurring during storage that
can lead to undesirable product speciations. Moreover, Patwardhan et al. recently demonstrated
that acetic acid promotes an oligomerization of phenolic monomers from lignin pyrolysis, thus
103
likely leading to pyrolytic lignin formation[5]. Additionally, abundance of organic acids
contributes to high oxygen content (typically 35-40 wt%) of bio-oil, thus lowering its higher
heating value (HHV). In particular, as acetic acid and propanoic acid contain 53 and 43 wt%
oxygen, respectively, the acids are considered one of the most highly oxygenated compounds
among bio-oil species. Thus, in order for bio-oil to be successfully upgraded into fuels and
chemicals, the removal of carboxylic acids from bio-oil would be a first step in the upgrading of
biofuels.
Ketonization is an ideal reaction for removing carboxylic acids as it takes two carboxylic
acids and converts them to ketone, carbon dioxide, and water, as follows:
→
It is noted that the reaction not only eliminates the acidic carboxylic acids and reduces oxygen
content (for instance, from acetic acid, 53 wt%, to acetone, 27 wt%), but also creates longer
chain molecules, ketone, via C-C coupling. Importantly, ketone is quite desirable molecule, since
the molecule can participate in multiple condensation reactions, such as aldol condensation and
alkylation that lead to sufficiently long C-C chain to be in conventional gasoline and diesel[6, 7].
Due to the great advantages of ketonization in bio-oil upgrading, a number of studies on
ketonization in biomass conversion with metal oxides have been recently conducted. However, a
mechanistic understanding of the ketonization is still debated. There have been two completely
different reaction mechanisms (bulk ketonization and surface ketonization) to describe
ketonization, depending on the metal oxide catalyst used. First, it is important to distinguish
between bulk ketonization, in which a thermal decomposition of metal carboxylate occurs to
release ketone, and surface ketonization, in which surface-catalyzed reaction (via α-hydrogen)
occurs to promote a ketone formation. It has been known that metal oxides with low lattice
104
energies (MgO, BaO, CaO, and SrO) promote an interaction between metal and organic acids to
form bulk carboxylate salts, releasing a ketone upon heating[8]. In contrast, surface-catalyzed
ketonization has been reported with metal oxides with high lattice energies, such as ZrO2, CeO2,
TiO2, and SnO2[9-11]. More recently, Snell et al. reported that ketonization of CeO2 proceeds via
either bulk or surface ketonization, depending on temperature. At low temperatures (150-300
°C), ketone is generated via a thermal decomposition of bulk carboxylate salts, while at high
temperatures (over 300 °C) ketonization proceeds on the catalyst surface by interaction of
carboxylic acid molecules with a surface.
As the ketonization reaction typically is performed at high temperature around 300-450
°C, it is not industrially practical to perform the reaction in the condensed phase bio-oil as high
pressure will be necessary. Moreover, once condensed, it is difficult to vaporize the bio-oil due
to its high content of thermally unstable compounds. Therefore, if the ketonization reaction is to
be used for bio-oil upgrading, it will need to be performed prior to condensation of pyrolysis
vapors. For bio-oil upgrading a number of studies on ketonization were performed. However, the
work has typically been conducted at high temperature and with active metal catalysts such as
CeO2 or CeZrOX[12, 13]. The reaction condition with active material would be undesirable as
they would likely promote a number of side reactions besides ketonization. Thus, it would be
preferred to remove/separate acids with inert material, and then to perform the ketonization
reaction prior to condensation of pyrolysis vapors so as to not expose the other bio-oil
compounds to the material at high temperatures.
In the current study, CaCO3, relatively inert material, was used to remove acids in
pyrolysis vapors and to convert them to ketone, carbon dioxide, and water. The performance of
the material was determined by a selectivity toward acid removal. In addition, we demonstrated a
105
successful regeneration of the material, and provided mechanical insights into the acid removal
and regeneration processes by characterizing the spent materials with XRD, XPS, and TGA.
Based on the mechanistic understandings, the material surface was modified to optimize acid
removal process.
Experimental
In the current work, all chemicals were purchased from Sigma-Aldrich unless otherwise
mentioned. Holocellulose, which is a native mixture of cellulose and hemicellulose, was
prepared according to the method shown in previous study[14].
Surface area of catalytic materials was measured using a Micromertitics ASAP 2020 with
a method of nitrogen physisorption at 77 K. Before the measurement, the sample was degassed
with a heating rate of 10 °C/min from room temperature to 350 °C and stayed at 350 °C for 4 h.
X-ray photoelectron spectroscopy (XPS) was performed using a Physical Electronics 5500
Multi-technique system. Al Kα radiation was used during the measurement. Gold and Cu was
used to calibrate the linearity of the energy scale. Using this shift calibration for the sample,
accurate energy scale was obtained. For etching experiment, bombardment of the sample by Ar+
ions at 4kV under 1 miroamp target current was used. The etching time was the length of time
the beam was on the sample. Crystallinity was determined by X-ray diffraction (XRD). A
Siemens D 500 X-ray diffractometer was used at room temperature with CuK radiation. All
specimens were scanned from 10 to 65 2θ with a step size of 0.05 degrees. TGA
(thermogravimetric analysis)-MS (mass spectrometry) analysis was performed by using a
Netzsch DSC/TGA-MS. During the analysis, mass change over temperature/time was recorded
and the identity of gases evolved during the heating process could be simultaneously determined
by the MS. During the measurement, the starting temperature of furnace was set at 80 °C with an
106
initial hold of 5 min. Then the furnace was heated up to 900 °C with a ramping rate of 20 K/min.
Both furnace temperature and actual sample temperature was recorded and in the current study
while only the sample temperature was used for plot. Mass fragment ions with m/z of 18, 44, and
58, representing water, CO2 and acetone, respectively, were chosen to be monitored during the
analysis.
The vapor phase acid removal from pyrolysis was investigated by using a Tandem microreactor system (Rx-3050 TR, Frontier Laboratories, Japan). Configuration of the microreactor
system is shown in Figure 1. The first reactor was used for fast pyrolysis. In the current study, a
temperature of 500 °C was used for all pyrolysis. The vapor resulted from the pyrolysis was
swept by continuous He flow to the second reactor, where acid removal occurred. The
temperature of the second reactor was tested at different temperatures, to tune the selectivity for
the acid removal. A fixed catalyst/adsorbent bed was packed in a quartz tube in the second
reactor. To prevent bypass flow, particle size of the bed was kept smaller than 1/10 of the tube
diameter, and the length of the bed was adjusted to be more than 4 times of the tube diameter.
Catalyst pellets were mixed with inert sand when bed length was not long enough. Both the
catalyst pellets and inert sand particles were sieved to 50 - 70 mesh size. Quartz wool was placed
at both ends of the quartz tube to immobilize the bed. For non-catalytic experiment, the second
reactor was empty and maintained at 300 °C. There are two interfaces in the micro-reactor
system, the first one of which is between the two reactors and the second one is between the
second reactor and GC system. The temperature for both interfaces was maintained at 300 °C to
minimize product condensation. The products coming out from the second reactor were online
analyzed by GC (7890A, Agilent Technologies, USA) equipped by three detectors. A three-way
splitter placed in front of columns enabled simultaneous analysis by the three detectors. The MS
107
was used for product identification. The flame ionization detector (FID) was used for
quantification of condensable products and the thermal conductivity detector (TCD) was used for
quantification of CO and CO2. Details for temperature ramping program, product identification
and calibration methods could be found in a previous publication[15]. Postreaction catalyst bed
was in situ regenerated in the microrector system by heating up the bed to a proper temperature
under He flow for 30 mins. Volatile products during the regeneration were identified by the MS
detector. In the current study, yield of products was calculated by using weight of product
divided by weight of reactant. The acetic acid removal was calculated by using the acetic acid
yield after the acid removal divided by its yield from non-catalytic pyrolysis. The loss of other
compounds was calculated by using the overall yield of condensable products (except for acetic
acid) after the acid removal divided by the over yield of these products from non-catalytic
pyrolysis.
Results and discussion
Product distribution for holocellulose pyrolysis
Holocellulose was obtained by delignification of cornstover, and was used as a feedstock
in the study. According to its compositional analysis in Table 1, delignification was successfully
performed so that only trace amount of lignin, 3 wt%, remained in the feedstock. The product
distribution of holocellulose pyrolysis without the sorbent is shown in Table S1. Non-sorbent
experiment was performed by using an empty quartz tube for the second reactor. A high mass
balance closure was achieved with 94 wt% and the yield of acetic acid was 9.2 wt%.
Material screening
8 different metal oxides/carbonates were tested in the tandem micro-reactor to evaluate
their performance in selective acid removal. As shown in Table 2, four materials (CaCO3, ZrO2,
108
MgCO3, and CaO) showed very high activity for removing acetic acid. However, these four
materials also removed considerable amounts of the other compounds. Among these materials,
CaCO3 showed apparently higher selectivity toward acid removal, and thus CaCO3 was selected
for acid removal in the following study.
The detailed product distribution after acid removal by fresh CaCO3 is shown in Table
S2. The diminishment in overall mass balance was due to the adsorbed acetic acid and “coking”
on the CaCO3. The CaCO3 bed turned to a dark brown color after the acid removal. Importantly,
carbon dioxide increased from 13.5 % (non-sorbent) to 18.4 % (after acid removal). We presume
that the adsorption of acetic acid on CaCO3 forms calcium acetate (CaAc), which would mean
carbon dioxide is released. By the stoichiometric relationship, approximately 3.4 wt% of carbon
dioxide would be generated when 9.2 wt% of acetic acid was reacted with CaCO3. When
including the amount of carbon dioxide produced during holocellulose pyrolysis, the overall
yield of carbon dioxide would be 16.5 wt%. The slightly higher yield of carbon dioxide of 18.4
wt% indicates decomposition of other compounds into carbon dioxide during acid removal. By
comparing Tables S1 and S2, significant decreases were observed for the yields of
glycolaldehyde (8.1 to 0.5 wt%), methyl glyoxal (1.3 to 0.1 wt%), acetol (3.0 to 1.1 wt%)
levoglucosan (4.1 to 2.5 wt%), and furfural (0.6 to 0.2 wt%). It should be noted that these
compounds have relatively higher oxygen content, which have been reported to have higher
reactivity in the presence of either acid or base catalysts.
Optimizing bed temperature and material loading
Reaction parameters, bed temperature and material loading, were varied to optimize the
selective acid removal. Two different temperatures, 250 and 300 C, were tested. The results
showed that 300 C gave a significantly higher degree of mass loss for the other compounds (up
109
to 83 wt ), accompanied by a only slight increase in acid removal relative to 250 C. This effect
was due to enhanced side reactions at higher temperature. A temperature lower than 250 °C was
not tested, since these temperatures would lead to condensation of heavy pyrolysis products such
as levoglucosan. As the loss of other compounds is undesirable, a temperature of 250 °C for the
bed was selected. Different loadings of CaCO3 (25, 50, and 100 mg) were tested at 250 °C. As
shown in Table 3, the acid removal at 25 mg was lower than that at higher loadings. The acid
removal at 100 mg was similar to that at 50 mg, but with a higher loss of other compounds. Thus,
50 mg of CaCO3 at 250 °C was selected as the basis condition for the remainder of the
experiments.
Stability testing
For stability testing, acid removal was performed with the same bed material of 50 mg at
250 °C for three consecutive pyrolysis events. As shown in Table 4, the acid removal
performance significantly decreased after each run. The fresh material showed a high degree (95
%) of acid removal; however, on the second run only 50 % of acetic acid was removed.
Furthermore, on the fourth run only a negligible amount of acid was removed. This suggests the
bed needs to be regenerated after each run to maintain acid removal activity. The rapid
deactivation along with a large material and feedstock ratio of 100:1 in the study leads to
hypothesis that the effective active sites for acid removal are located on surface of the material
with pores possibly blocked by an intermediate formed during acid removal.
Insights into acid removal with CaCO3
The postreaction CaCO3 was subjected to XRD analysis, the result of which was
compared with the fresh CaCO3 as shown in Figure 2. No bulk phase change was observed,
suggesting the bulk phase of the postreaction material was still CaCO3. Interestingly, as shown in
110
Table 4, this postreaction material showed little effect in acid removal. Therefore, the acid
removal process was likely an adsorption process instead of absorption.
To verify the hypothesis of adsorption, XPS analysis on the postreaction CaCO3 was
performed, the result of which was compared with the fresh CaCO3 and CaAc, as shown in
Figure 3. For CaCO3, the C1s peak at 289.3 eV represents the carbonate carbon in CaCO3 while
the C1s peak at 284.9 eV might be assigned to adventitious carbon contamination[16]. Based on
our knowledge, the C1s XPS spectrum for calcium acetate has not been reported in literatures.
To assign the C1s peaks for calcium acetate, literatures containing the C1s XPS spectra of
materials with different carbon functional groups were referred [17-22]. In the current study, the
C1s peak at 287.6 eV represents the carboxylate carbon in CaAc while the C1s peak at 284 eV
represents the methyl carbon in CaAc. Comparing to fresh CaCO3, a peak shifting towards CaAc
was observed for the postreaction CaCO3, from 289.3 eV to 288.3 eV. Moreover, a peak at 283.7
eV, representing the methyl carbon, was observed in the postreaction CaCO3. The peak shifting
and formation for the postreaction CaCO3 suggested a formation of acetate group on the surface
of the material during the acid removal.
To further verify the adsorption hypothesis, etching experiments were performed for the
postreaction CaCO3. Figure 4 shows the XPS spectra of postreaction CaCO3 with different
etching degree. It appeared that as the etching intensity increased, the C1s peak on the left side
gradually shifted from the carboxylate carbon to the carbonate carbon. Simultaneously, the
intensity of the right peak, which represents the methyl carbon, gradually decreased. This
demonstrates the acetate group formed during the acid removal was located on the surface of
CaCO3, which could be cleaned by the etching. Also, the etching removed the adventitious
carbon contamination as the relevant peak disappeared after extensive etching. Therefore, the
111
postreaction material is converted to clean CaCO3 after enough etching. The etching time is
proportional to the depth of the materials removed. In the current experiment, the etching rate is
about 1 nanometer/minute. Therefore, an estimation of the thickness of acetate layer formed
during the acid removal could be obtained based on Figure 4. The molecular size of the CaAc
could be estimated based on its true density or empirical formula[23], which suggests a value of
0.69 nm or 0.67 nm, respectively. Therefore, if a monolayer of acetate molecule was formed on
the surface after the acid removal, 42 s or 40 s was needed to remove it. By examining the left
C1s peak in Figure 4, an etching time of 80 s led to the peak shifting back to CaCO3 while an
etching time of 20 s was not enough to complete the peak shifting. The result suggested the
monolayer acetate might be formed on the surface of CaCO3 during the acid removal.
To further verify the monolayer adsorption model, results of multiple acid removal using
the same CaCO3 bed were used as shown in Table 4. The number of CaCO3 molecules on the
surface of the catalyst bed was calculated based on surface area of the CaCO3 and an estimated
cross section area of the CaCO3 molecule which was calculated based on its true density and
molecular weight. Given the acid removal stoichiometry, the number of available surface site on
the catalyst bed was compared to the number of removed acetic acid molecules which was
calculated based on Table 4. The calculation shows that under the circumstance of monolayer
adsorption, maximum amount of acetic acid could be adsorbed on the catalyst bed is 1.17E-06
mole, which is slightly larger than the actual amount of acid removed after the three runs, 1.07E06 mole. Little acid removal was observed after the 3rd run (<1%), which was probably due to
little availability of surface sites leading to low adsorption rate. This calculation validated the
monolayer adsorption hypothesis.
112
Material regeneration
The characterization of the postreaction CaCO3 suggested the formation of acetate groups
on the surface of CaCO3 during the acid removal. In order to choose the proper regeneration
temperature, TGA-MS was used for the postreaction CaCO3. The postreaction CaCO3 was
obtained by performing two runs of holocellulose pyrolysis over the fresh CaCO3. 56 µg of
acetic acid was removed during the two runs. Molecular ion tracking for acetone and CO2
suggested the regeneration started from 350 °C and ended at 550 °C. TGA-MS for CaAc showed
its decomposition to CaCO3 over a very similar temperature range as for the postreaction CaCO3
regeneration, as suggested in Figure S1. Due to a very small CaAc content in the postreaction
CaCO3 based on the aforementioned calculation, a magnified TGA curve between 300 °C and
550 °C is necessary to observe the degree of regeneration over temperature, as shown in Figure
5(b). Figure 5(b) showed a weight loss of 29.6 µg from 350 °C to 550 °C, which is only slightly
larger than 27.0 µg, the theoretical amount of acetone could be formed based on the 56 µg of
adsorbed acetic acid. The difference should be the amount of CO2 formed during the
decomposition of CaAc between 350 °C and 550 °C show in Figure 5(d). Further increase in
temperature would lead to decomposition of CaCO3 to CaO, as suggested by large amount of
CO2 was formed after 550 °C in Figure 5(d). TGA-MS for fresh CaCO3 was also shown in
Figure S2 as a reference. The TGA-MS results suggested the CaCO3 regeneration started from
350 °C and completed at 550 °C.
Two postreaction CaCO3 beds were in situ regenerated at 550 °C and 700 °C for 30 mins,
respectively. As shown in Figure S3, acetone was formed during the regeneration, which
suggested the surface acetate group decomposed into CaCO3 during the regeneration. These two
materials were then subjected to XRD analysis. Figure 6 showed the CaCO3 regenerated at 550
113
°C only had characteristic peaks for CaCO3 while regeneration at 700 °C leads to presence of
characteristic peaks for both CaCO3 and CaO. This is consistent with TGA-MS results that
CaCO3 decomposition into CaO starting after 550 °C. XRD calculation showed the content of
CaO is about 10 wt% in CaCO3 regenerated at 700 °C. Both regenerated CaCO3 beds (550 °C
and 700 °C) were then used for acid removal at the same experiment conditions as the fresh
CaCO3. As shown in Table 5, similar degree of acid removal and loss of other compounds was
observed between the fresh and 550 °C regenerated CaCO3, which suggested the complete
regeneration of CaCO3 at 550 °C. On the contrary, the 700 °C regenerated CaCO3 showed much
larger degree of loss of other compounds due to the presence of CaO, which was consistent with
the result in Table 2. The complete regeneration at 550 °C was further verified by XPS analysis
in the following section. The complete product distribution by using fresh CaCO3 bed and the
bed regenerated at 550 °C was shown in Table S2.
Optimizing the selectivity by surface modification
As shown in Table 2, the fresh CaCO3 had the highest selectivity towards acid removal.
However, about 57 % loss of other compounds was observed even for this case. To optimize the
selectivity, partial passivation on the surface of CaCO3 was performed. Acetate functional group
was selected due to weaker basicity of CaAc and its formation during the acid removal as
suggested by XPS and TGA studies. The less loss of other compounds might be resulted from
the weaker basicity of CaAc compared to CaCO3, as suggested by previous study that the
stronger base was more active in catalytic conversion of cellulose pyrolysis derived oxygenates.
A comparison experiment between the carbonate group and acetate group was performed by
using a proper amount of CaAc (3mg) to make the catalyst/adsorbent bed, which had the same
surface area as the 50 mg of CaCO3. Inert sand particles were mixed with CaAc to maintain the
114
same bed length as the 50 mg CaCO3. Under the same experimental conditions as Table 4, less
than 6 % of loss of other compounds was observed for CaAc although no acid removal was
observed as well. Therefore, incorporation of acetate group on the surface of CaCO3 might be a
route to mitigate the loss of other compounds. A straightforward method to perform the surface
modification is to choose a moderate temperature between 350 °C and 550 °C, as suggested by
the TGA-MS analysis, to partially regenerate the postreaction CaCO3, leading to residue acetate
groups on the surface.
Three postreaction CaCO3 beds, which were the same as the one in Figure 3, underwent
the partial regeneration at 440 °C, 420 °C, and 400 °C, respectively. The partial regeneration was
performed using the same protocol as the regeneration at 550 °C, except for the different
regeneration temperatures. The XPS spectra for the regenerated CaCO3 was shown in Figure 7.
XPS spectra of CaCO3 and CaAc were also included as references. For the left peak of the
postreaction CaCO3, regeneration resulted in a shift towards carbonate carbon from carboxylate
carbon. Higher regeneration temperature led to a larger shift for this peak. This is consistent with
TGA-MS results that the regeneration continued until 550 °C. Similar shifting behavior was
observed for the right peak, representing removal of the methyl carbon in acetate during the
regeneration. Complete regeneration would lead it to shifting back to the adventitious carbon
contaminations in the fresh CaCO3. The partial regenerated at 440 °C, as suggested by Figure 7,
is consistent with TGA-MS results which showed formation of acetone was not completed at 440
°C.
It is interesting to provide insights into the peak at 288.3 eV and 288.5 eV for
postreaction CaCO3 and regenerated CaCO3 at 440 °C, respectively. These two peaks were
located between carboxylate peak and carbonate peak. Based on the monolayer adsorption
115
calculation, about 92% of CaCO3 surface site already adsorbed acetic acid for the postreaction
CaCO3. Some of these sites were regenerated after regeneration at 440 °C. Similar phenomenon
was observed in previous work from our group about ceria catalyzed ketonization of acetic acid.
Under intermediate temperature, ceria restructuring was observed after ketonization, which was
assigned to a metal-carboxylate intermediate since the restructuring did not match the structure
of ceria nor cerium acetate. It should be noted that this restructuring was not observed at higher
ketonization temperature which led to a complete catalysis cycle. It was also reported that under
the intermediate temperature, both the cerium acetate hydrate and cerium carbonate hydrate
would form the same metal-carboxylate intermediate as the postreaction CeO2. This, combining
the results from TPD-MS for cerium acetate hydrate and postreaction CeO2, suggested such a
metal-carboxylate intermediate was possible in a form of metal-acetate-carbonate. Similarly, in
the current work the peak at 288.5 eV for the partially regenerated CaCO3, which was located
between CaCO3 and CaAc, might be assigned to the metal-carboxylate intermediate as
previously proposed[24].
The partially regenerated catalysts beds were used for acid removal, the results of
which were shown in Table 8. The complete product distribution was shown in Table S3. The
unaccounted mass balance in Table S3, which is proportional to the loss of other compounds,
was largely due to coke formation and coking associated water formation as the postreaction
catalyst bed was turned to black. It appeared that the selectivity for acid removal could be tuned
by using different regeneration temperature. The catalyst bed regenerated at 440 °C showed only
slightly decrease in acid removal compared to fresh catalyst but decreased the loss of other
compounds from 57% to 20%. The physisorption analysis showed the BET surface area for this
regenerated CaCO3 is 1.19 m2/g which is very similar to 1.32 m2/g as the value for the fresh
116
CaCO3. This suggested the change in selectivity was not derived from the change in number of
sites on the surface but the change in nature of these sites. Also observed in Table 8 was the
decreased degree of acid removal and loss of other compounds as the decreasing regeneration
temperature. This was possibly resulted from the different ratio of carbonate/acetate in the
intermediate, as suggested by the TGA-MS and XPS. Lower regeneration temperature resulted in
lower degree of acetate decomposition. The resulted material had less acid removal effect as well
as less loss of other compounds. This could be explained by the inertness of CaAc during the
acid removal. The amount of CO2 formed by the reaction between acetic acid and CaCO3, as
shown in Table S3, increased proportional to the degree of acid removal. Therefore, depending
on the emphasis on acid removal against preventing product loss in a specific application, proper
catalytic material could be chosen among these partially regenerated CaCO3 with different
carbonate/acetate ratio on the surface. This ratio could be tuned by temperature used in the heat
treatment for the postreaction CaCO3.
The reusability of the regenerated catalyst bed at 440 °C was tested. As shown in Figure
8, up to fourth cycles no apparent change was observed for degree of acid removal and loss of
other compounds. Moreover, the XPS and physisorption was performed for regenerated catalyst
after the fourth cycle. Similar location of C1s peaks was observed as the catalyst after 1st
regeneration at 440 °C. Negligible change on the BET surface area was observed as well. This
suggested the ratio of carbonate/acetate in the metal-carboxylate intermediate could be preserved
after each regeneration as long as the same regeneration temperature was used.
Conclusions
Gas phase organic acid removal for holocellulose pyrolysis vapor was studied in a
Tandem microreactor system. A GC system equipped by three detectors was used for online
117
product analysis, from which high mass balance closure for pyrolysis products was obtained.
Typical catalysts for ketonization, such as metal oxide and metal carbonate, were tested, among
which CaCO3 showed the best selectivity towards the acid removal. In situ catalyst regeneration
was performed by heat treatment on the postreaction catalyst bed at proper temperature. During
the regeneration, ketone was formed from the adsorbed acid. Extensive characterization
techniques were used for the fresh, postreaction, and regenerated CaCO3, including XPS, BET,
XRD, TGA-MS. It was suggested that the acid removal occurred only on the surface of CaCO3.
Theoretical calculation suggested the acid removal was probably a monolayer adsorption
process. Surface modification on CaCO3 was performed by heating the postreaction material at
intermediate temperatures. The surface modified CaCO3 materials were capable of mitigating
side reactions during the acid removal. Characterization suggested a formation of a metalcarboxylate intermediate after the surface modification, which consists both acetate and
carbonate groups. The incorporated acetate group was proven to be inert for the pyrolysis vapors.
As such, the selectivity towards acid removal was largely improved after the surface
modification. Furthermore, the selectivity could be tuned by the ratio of acetate/carbonate in the
metal-carboxylate intermediate, which was controlled by temperature for the modification. By
using the surface modified CaCO3, about 95% acid could be removed at the expense of 20% loss
of other compounds. Both fresh CaCO3 and the surface modified CaCO3 showed satisfactory
reusability.
References
1.
Stevens, C. and R.C. Brown, Thermochemical processing of biomass: conversion into
fuels, chemicals and power. Vol. 12. 2011: John Wiley & Sons.
2.
Huber, G.W., S. Iborra, and A. Corma, Synthesis of transportation fuels from biomass:
chemistry, catalysts, and engineering. Chemical reviews, 2006. 106(9): p. 4044-4098.
118
3.
Mohan, D., C.U. Pittman, and P.H. Steele, Pyrolysis of Wood/Biomass for Bio-oil: A
Critical Review. Energy & Fuels, 2006. 20(3): p. 848-889.
4.
Choi, Y.S., et al., Detailed Characterization of Red Oak-Derived Pyrolysis Oil;
Integrated use of GC, HPLC, IC, GPC and Karl-Fischer. Journal of Analytical and
Applied Pyrolysis, 2014.
5.
Patwardhan, P.R., R.C. Brown, and B.H. Shanks, Understanding the fast pyrolysis of
lignin. ChemSusChem, 2011. 4(11): p. 1629-1636.
6.
Zapata, P.A., et al., Condensation/hydrogenation of biomass-derived oxygenates in
water/oil emulsions stabilized by nanohybrid catalysts. Topics in Catalysis, 2012. 55(12): p. 38-52.
7.
Nie, L. and D.E. Resasco, Improving carbon retention in biomass conversion by
alkylation of phenolics with small oxygenates. Applied Catalysis A: General, 2012. 447:
p. 14-21.
8.
Lafer, L., V. Yakerson, and A. Rubinshtein, Infrared spectra of catalysts and adsorbed
molecules. Russian Chemical Bulletin, 1967. 16(11): p. 2312-2318.
9.
Pestman, R., et al., Reactions of carboxylic acids on oxides: 1. Selective hydrogenation of
acetic acid to acetaldehyde. Journal of catalysis, 1997. 168(2): p. 255-264.
10.
Pestman, R., et al., Reactions of carboxylic acids on oxides: 2. Bimolecular reaction of
aliphatic acids to ketones. Journal of Catalysis, 1997. 168(2): p. 265-272.
11.
Pestman, R., et al., The formation of ketones and aldehydes from carboxylic acids,
structure-activity relationship for two competitive reactions. Journal of Molecular
Catalysis A: Chemical, 1995. 103(3): p. 175-180.
12.
Gärtner, C.A., et al., Catalytic Upgrading of Bio‐Oils by Ketonization. ChemSusChem,
2009. 2(12): p. 1121-1124.
13.
Deng, L., Y. Fu, and Q.-X. Guo, Upgraded acidic components of bio-oil through
catalytic ketonic condensation. Energy & Fuels, 2008. 23(1): p. 564-568.
14.
Zhang, J., Fast pyrolysis behavior of different celluloses and lignocellulosic biopolymer
interaction during fast pyrolysis. 2012.
15.
Zhang, J., M.W. Nolte, and B.H. Shanks, Investigation of Primary Reactions and
Secondary Effects from the Pyrolysis of Different Celluloses. ACS Sustainable Chemistry
& Engineering, 2014. 2(12): p. 2820-2830.
16.
Boughriet, A., et al., Generation of uranyl/carbonate/hydroxide ‘‘coatings’’at the vaterite
surface. Physical Chemistry Chemical Physics, 2000. 2(5): p. 1059-1068.
119
17.
Feng, R., et al., In situ synthesis of poly (ethylene terephthalate)/graphene composites
using a catalyst supported on graphite oxide. Journal of Materials Chemistry, 2011.
21(11): p. 3931-3939.
18.
Hawaldar, R., et al., Large-area high-throughput synthesis of monolayer graphene sheet
by hot filament thermal chemical vapor deposition. Scientific reports, 2012. 2.
19.
Tu, Q., et al., Effects of surface charges of graphene oxide on neuronal outgrowth and
branching. Analyst, 2014. 139(1): p. 105-115.
20.
Stankovich, S., et al., Synthesis of graphene-based nanosheets via chemical reduction of
exfoliated graphite oxide. Carbon, 2007. 45(7): p. 1558-1565.
21.
Kovtyukhova, N.I., et al., Individual single-walled nanotubes and hydrogels made by
oxidative exfoliation of carbon nanotube ropes. Journal of the American Chemical
Society, 2003. 125(32): p. 9761-9769.
22.
Yang, D., et al., Chemical analysis of graphene oxide films after heat and chemical
treatments by X-ray photoelectron and Micro-Raman spectroscopy. Carbon, 2009. 47(1):
p. 145-152.
23.
Wang, H. and M. Frenklach, Transport properties of polycyclic aromatic hydrocarbons
for flame modeling. Combustion and flame, 1994. 96(1): p. 163-170.
24.
Snell, R.W. and B.H. Shanks, Insights into the Ceria-Catalyzed Ketonization Reaction
for Biofuels Applications. ACS Catalysis, 2013. 3(4): p. 783-789.
120
Tables and figures
Figure 2. Configuration of Tandem micro-reactor coupled with a GC system.
Table 2. Holocellulose composition
Components
Glucan
Xylan
Galactan
Arabinan
Mannan
Lignin
Ash
Total
Composition, wt%
45.8
27.6
2.6
5.0
1.1
3.0
3.6
88.7
121
Table 3. Material screening for a selective acid removal
Acetic acid removal, wt%
Mass loss of other cmpds, wt%
CaCO3
98.1
56.6
ZrO2
100.0
93.2
MgCO3
100.0
87.6
CaO
100.0
85.5
Ce2(CO3)3
79.7
55.5
CeO2
61.6
54.0
TiO2
47.1
48.8
MgO
0
7.8
Reaction conditions: pyrolysis temperature = 500 °C; material temperature = 250 °C; reactant
loading = 0.5 mg; material loading = 50 mg
Table 4. Selectivity toward acid removal as a function of CaCO3 loading
Acetic acid removal, wt%
Mass loss of other cmpds, wt%
25 mg
59.1
21.6
50 mg
98.1
56.6
100 mg
98.9
83.8
Reaction conditions: pyrolysis temperature = 500 °C; material temperature = 250 °C; reactant
loading = 0.5 mg
Table 5. Stability test of CaCO3 for acid removal
Acetic acid removal, wt%
Mass loss of other cmpds, wt%
1st run
98.1
56.6
nd
2 run
48.3
18.9
3rd run
9.1
2.0
th
4 run
<1
0
Reaction conditions: pyrolysis temperature = 500 °C; material temperature = 250 °C; reactant
loading = 0.5 mg; material loading = 50 mg
122
Intensity, CPS
Fresh CaCO₃
0
Postreaction CaCO₃
10
20
30
40
50
60
70
2θ
Figure 3. XRD patterns for fresh CaCO3 and postreaction CaCO3. The postreaction CaCO3 was
obtained after three runs of holocellulose pyrolysis vapor over fresh CaCO3. (Reaction
conditions: pyrolysis temperature = 500 °C; acid removal temperature = 250 °C; holocellulose
loading = 0.5 mg; CaCO3 loading = 50 mg)
Postreaction CaCO₃
Fresh CaCO₃
Fresh CaAc
5000
3000
2000
Intensity, CPS
4000
1000
0
295
292
289
286
Binding energy, eV
283
280
Figure 4. XPS spectra (C1s area) for fresh CaCO3, fresh CaAc, and postreaction CaCO3. The
postreaction CaCO3 was obtained after three runs of holocellulose pyrolysis vapor over fresh
CaCO3. (Reaction conditions: pyrolysis temperature = 500 °C; acid removal temperature =
250 °C; holocellulose loading = 0.5 mg; CaCO3 loading = 50 mg)
123
6000
20 sec etching
260 sec etching
Fresh CaAc
5000
4000
3000
2000
Intensity, CPS
Postreaction CaCO₃, no etching
80 sec etching
Fresh CaCO₃
1000
0
295
292
289
286
Binding energy, eV
283
280
Figure 5. XPS spectra (C1s area) for fresh CaCO3, fresh CaAc, postreaction CaCO3 and
postreaction CaCO3 with different etching degree. Same postreaction CaCO3 was used as Figure
3.
124
(a) Weight loss
(c) m/z 58
1E-12
0.00E+00
Ion current, A
Sample weight, mg
4.00E+00
-4.00E+00
-8.00E+00
-1.20E+01
-1.60E+01
1E-13
0 150 300 450 600 750 900
Sample temperature, °C
1E-08 0
(b) Weight loss
(d) m/z 44
1.00E-02
1E-09
0.00E+00
Ion current, A
Sample weight, mg
200 400 600 800
Sample temperature, °C
-1.00E-02
-2.00E-02
-3.00E-02
1E-10
1E-11
300
400
500
Sample temperature, °C
0
200 400 600 800
Sample temperature, °C
Figure 6. TGA-MS results for postreaction CaCO3. (a) sample weight loss over full temperature
range, (b) sample weight loss from 300 °C to 550 °C, (c) mass fragment ion from acetone
monitored by MS, (d) mass fragment ion from CO2 monitored by MS.
125
Intensity, CPS
CaCO₃ regenerated 700
°C
0
10
20
30
40
50
60
70
2θ
Figure 7. XRD patterns for CaCO3 regenerated at different temperature. Triangle: peaks for
CaCO3; round, peaks for CaO.
Table 6. Degree of acid removal and loss of other compounds for the fresh and regenerated
CaCO3
Fresh
Regenerated
Regenerated
CaCO3
at 550 °C
at 700 °C
Acetic acid removala
98.1
97.5
98.0
Loss of other compoundsb
56.5
46.8
72.0
126
5000
Fresh CaCO₃
CaCO₃ regenerated 440 °C
CaCO₃ regenerated 550 °C
Fresh CaAc
Postreaction CaCO₃
4000
3000
2000
1000
0
295
292
289
286
283
280
Figure 8. XPS spectra (C1s area) for fresh CaCO3, fresh CaAc, postreaction CaCO3 and
postreaction CaCO3 regenerated at different temperatures. The postreaction CaCO3 is the same
one as in Figure 3.
127
Table 8. Major product distribution after the acid removal over different catalyst beds
Products
Fresh
Regeneration Regeneration Regeneration
CaCO3 at 440 °C
at 420 °C
at 400 °C
1.12
1.21
1.22
1.28
0.13
0.38
0.57
0.71
0.47
3.27
4.47
5.28
0.18
0.72
1.45
2.11
1.09
4.28
3.94
3.70
0.12
0.49
0.69
0.74
0.23
0.45
0.63
0.67
0.19
0.63
0.69
0.71
0.25
0.49
0.59
0.66
2.11
3.05
3.62
3.91
0.73
0.94
1.13
1.16
Acetaldehyde
Methyl glyoxal
Glycolaldehyde
Acetic acid
Acetol
Acetic acid, methyl ester
MW 86
MW 102
2-Furaldehyde
DAXP 1
2(5H)Furanone
2(3H)-Furanone, dihydro-4hydroxy0.78
1.02
1.08
1.16
1,4;3,6-Dianhydro-α-Dglucopyranose
0.44
0.36
0.34
0.37
5-(Hydroxymethyl)-2-furaldehyde
0.71
0.51
0.47
0.50
Levoglucosan
2.49
3.17
2.79
3.00
CO
4.27
4.71
4.49
4.14
CO2
18.38
17.48
16.95
16.45
a
Loss of other compounds
56.56
20.31
16.33
10.34
Acid removalb
98.05
94.69
84.12
76.93
CO
4.27
4.71
4.49
4.14
CO2
18.38
16.68
14.95
14.45
a, in unit of percentage; b, in unit of percentage. Reaction conditions: pyrolysis temperature =
500 °C; material temperature = 250 °C; reactant loading = 0.5 mg; material loading = 50 mg
128
Acid removal, %
Loss of other compounds, %
100
80
60
40
20
0
1st
2nd
3rd
4th
regeneration at regeneration at regeneration at regeneration at
440 °C
440 °C
440 °C
440 °C
Figure 9. Reusability test for CaCO3 regenerated at 440 C (Reaction conditions: pyrolysis
temperature = 500 °C; material temperature = 250 °C; reactant loading = 0.5 mg; material
loading = 50 mg)
129
Supporting information
Table S1. Product distribution from holocellulose pyrolysis without adsorbent
Compound
Yield, wt%
Formaldehyde
0.21±0.03
Acetaldehyde
1.03±0.03
Furan
0.04±0.00
2-propanal
0.21±0.01
Acetone
0.15±0.01
Methyl glyoxal
1.35±0.03
2-methyl furan
0.04±0.00
2-pentanone
0.17±0.04
Glycolaldehyde
8.09±3.63
Acetic acid
9.16±0.45
Acetol
3.03±0.23
Methyl acetate
0.86±0.26
MW 86
0.77±0.04
MW 102
0.67±0.05
2-furaldehyde
0.60±0.05
3-furan methanol
0.04±0.00
DAXP 1
2.87±0.04
DAXP 2
0.33±0.01
2(5H)-furanone
1.24±0.02
Other DAXP 2
0.24±0.01
2-hydroxy-3-methyl-2-cyclopenten-10.21±0.01
one
Dihydro-4-hydroxy2(3H)-furanone
0.94±0.01
Dihydro-6-methyl-2H-pyran-3(4H)-one
0.41±0.04
1,4;3,6-dihydro-α-D-glucopyranose
0.60±0.11
AXP
0.07±0.00
5-(hydroxymethyl)-2-furaldehyde
0.67±0.04
Dianhydro glucopyranose
0.28±0.00
Other AXP
0.34±0.05
Levoglucosan
4.12±0.06
Carbon monoxide
4.50±0.11
Carbon dioxide
13.45±0.63
Char
16.59±0.26
Water
21.70
Total
94.96
Reaction conditions: pyrolysis temperature = 500 °C; reactant loading = 0.5 mg
130
(a) Weight loss
(b) Mass fragment ion
1E-08
1E-09
0.00E+00
Ion Current, A
Sample weight, mg
1.00E+01
-1.00E+01
-2.00E+01
1E-10
m/z 18
m/z 44
m/z 58
1E-11
1E-12
-3.00E+01
1E-13
0 150 300 450 600 750 900
Sample temperature, °C
0
150 300 450 600 750 900
Sample temperature, °C
Figure S1. TGA-MS results for CaAc. (a) weight loss over temperature, (b) mass fragment ion
monitored by MS over temperature.
(a) Weight loss
0.0000001
(b) Mass fragment ion
4.00E+00
1E-08
1E-09
Ion Current, A
Sample weight, mg
0.00E+00
m/z 18
m/z 44
m/z 58
-4.00E+00
-8.00E+00
-1.20E+01
1E-10
1E-11
1E-12
-1.60E+01
1E-13
0 150 300 450 600 750 900
Sample temperature, °C
0
150 300 450 600 750 900
Sample temperature, °C
Figure S2. TGA-MS results for CaCO3. (a) weight loss over temperature, (b) mass fragment ion
monitored by MS over temperature.
131
Figure S3. Mass spectrum during the regeneration of CaCO3 at 550 °C
132
Table S2. Complete product distribution for holocellulose pyrolysis vapors after the acid removal
over fresh CaCO3 and CaCO3 regenerated at 550 °C.
Products
Formaldehyde
Acetaldehyde
Furan
2-Propanal
Acetone
Methyl glyoxal
2-Methyl furan
2-Pentanone
Glycolaldehyde
Acetic acid
Acetol
Acetic acid, methyl ester
MW 86
MW 102
2-Furaldehyde
3-Furan methanol
DAXP 1
DAXP 2
2(5H)Furanone
Other DAXP 2
2-Hydroxy-3-methyl-2-cyclopenten-1one
2(3H)-Furanone, dihydro-4-hydroxy2H-Pyran-3(4H)-one, dihydro-6-methyl1,4;3,6-Dianhydro-α-D-glucopyranose
AXP
5-(Hydroxymethyl)-2-furaldehyde
Dianhydro glucopyranose
Other AXP
Levoglucosan
CO
CO2a
CO2b
CO2c
Char
Water from pyrolysis
Adsorbed acetic acid
Summation of products from
holocellulosed
Fresh CaCO3
0.17
1.12
0.04
0.13
0.14
0.13
0.06
0.08
0.47
0.18
1.09
0.12
0.23
0.19
0.25
0.01
2.11
0.17
0.73
0.03
CaCO3 regenerated
at 550 °C
0.23
1.23
0.05
0.21
0.19
0.25
0.06
0.18
2.30
0.23
3.70
0.44
0.38
0.48
0.37
0.02
0.59
0.09
0.51
0.03
0.21
0.78
0.26
0.44
0.11
0.71
0.27
0.29
2.49
4.27
13.45
3.29
1.64
16.59
21.70
8.98
0.22
0.53
0.08
0.39
0.09
0.32
0.10
0.23
2.44
4.50
13.45
3.28
0.87
16.59
21.70
8.93
79.64
81.98
133
Table S12. (continued)
Products
Fresh CaCO3
CaCO3 regenerated at 550 °C
AA removale
98.05
97.52
Loss of other compoundsf
56.56
46.82
a, from pyrolysis; b, from CaCO3; c, from sides reactions; d, summation of all products and
adsorbed acid except for CO2 from CaCO3; e, in unit of percentage; f, in unit of percentage.
Reaction conditions: pyrolysis temperature = 500 °C; material temperature = 250 °C; reactant
loading = 0.5 mg; material loading = 50 mg
134
Table S3. Complete product distribution for holocellulose pyrolysis vapors after the acid removal
over different catalyst beds.
Products
Formaldehyde
Acetaldehyde
Furan
2-Propanal
Acetone
Methyl glyoxal
2-Methyl furan
2-Pentanone
Glycolaldehyde
Acetic acid
Acetol
Acetic acid, methyl ester
MW 86
MW 102
2-Furaldehyde
3-Furan methanol
DAXP 1
DAXP 2
2(5H)Furanone
Other DAXP 2
2-Hydroxy-3-methyl-2cyclopenten-1-one
2(3H)-Furanone, dihydro-4hydroxy2H-Pyran-3(4H)-one, dihydro-6methyl1,4;3,6-Dianhydro-α-Dglucopyranose
AXP
5-(Hydroxymethyl)-2-furaldehyde
Dianhydro glucopyranose
Other AXP
Levoglucosan
CO
CO2a
CO2b
CO2c
Char
Water from pyrolysis
Fresh
CaCO3
0.17
1.12
0.04
0.13
0.14
0.13
0.06
0.08
0.47
0.18
1.09
0.12
0.23
0.19
0.25
0.01
2.11
0.17
0.73
0.03
Regeneration Regeneration Regeneration
at 440 °C
at 420 °C
at 400 °C
0.26
0.23
0.25
1.21
1.22
1.28
0.06
0.06
0.07
0.25
0.25
0.24
0.21
0.20
0.21
0.38
0.57
0.71
0.04
0.05
0.06
0.20
0.20
0.20
3.27
4.47
5.28
0.72
1.45
2.11
4.28
3.94
3.70
0.49
0.69
0.74
0.45
0.63
0.67
0.63
0.69
0.71
0.49
0.59
0.66
0.03
0.04
0.04
3.05
3.62
3.91
0.32
0.21
0.23
0.94
1.13
1.16
0.07
0.11
0.16
0.21
0.23
0.23
0.25
0.78
1.02
1.08
1.16
0.26
0.69
0.61
0.65
0.44
0.11
0.71
0.27
0.29
2.49
4.27
13.45
3.29
1.64
16.59
21.70
0.36
0.11
0.51
0.43
0.27
3.17
4.71
13.45
3.09
0.94
16.59
21.70
0.34
0.08
0.47
0.10
0.12
2.79
4.49
13.45
2.83
0.68
16.59
21.70
0.37
0.08
0.50
0.11
0.13
3.00
4.14
13.45
2.58
0.42
16.59
21.70
135
Table S3. (continued)
Products
Fresh
Regeneration
Regeneration
Regeneration
CaCO3
at 440 °C
at 420 °C
at 400 °C
8.98
8.44
7.71
7.05
79.64
89.96
90.80
91.96
Adsorbed acetic acid
Summation of products
from holocellulosed
Acid removale
98.05
94.69
84.12
76.93
f
Loss of other compounds
56.56
20.31
16.33
10.34
a, from pyrolysis; b, from CaCO3; c, from sides reactions; d, summation of all products and
adsorbed acid except for CO2 from CaCO3; e, in unit of percentage; f, in unit of percentage.
Reaction conditions: pyrolysis temperature = 500 °C; material temperature = 250 °C; reactant
loading = 0.5 mg; material loading = 50 mg
136
CHAPTER 6
GENERAL CONCLUSION
The research work presented in this dissertation focuses on a manipulation of reaction
pathways for selective bio-oil compositions, which is important in designing the downstream
catalytic processes for fuels and chemicals. Firstly, reaction pathways which govern product
speciations were investigated for lignin model compounds. Contrary to conventional belief that
free radical reaction is predominant in lignin pyrolysis, this study suggests that pericyclic and
free radical reactions dominantly occur in β-O-4 and α-O-4 typed model compounds,
respectively. Secondly, in situ catalytic fast pyrolysis was performed with catalysts having
acidity or basicity, to determine catalytic impacts to the bio-oil compositions. Full
characterization of the resulting bio-oils revealed that a significant amount of organics in bio-oil
was sacrificed during CFP, due to huge consumptions of C and H with catalysts. Thus, low total
energy recovered with CFP, compared to thermal pyrolysis. Finally, vapor product distribution
was manipulated by adsorbing acetic acid from pyrolysis vapors with CaCO3. By performing
extensive material analysis on postreaction CaCO3, insights of acid removal and material
regeneration were provided.